diff --git a/internal/cgen/base/image-impl.c b/internal/cgen/base/image-impl.c index 59217088..61ebd14a 100644 --- a/internal/cgen/base/image-impl.c +++ b/internal/cgen/base/image-impl.c @@ -21,6 +21,52 @@ const uint32_t wuffs_base__pixel_format__bits_per_channel[16] = { 0x08, 0x0A, 0x0C, 0x10, 0x18, 0x20, 0x30, 0x40, }; +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 src) { + size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + + size_t n = len4; + while (n--) { + uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s); + uint32_t b5 = 0x1F & (argb >> (8 - 5)); + uint32_t g6 = 0x3F & (argb >> (16 - 6)); + uint32_t r5 = 0x1F & (argb >> (24 - 5)); + wuffs_base__store_u32le__no_bounds_check( + d, (r5 << 11) | (g6 << 5) | (b5 << 0)); + s += 4; + d += 4; + } + return len4 * 4; +} + +static uint64_t // +wuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 src) { + size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + + size_t n = len4; + while (n--) { + uint8_t b0 = s[0]; + uint8_t b1 = s[1]; + uint8_t b2 = s[2]; + uint8_t b3 = s[3]; + d[0] = b2; + d[1] = b1; + d[2] = b0; + d[3] = b3; + s += 4; + d += 4; + } + return len4 * 4; +} + static inline uint32_t // wuffs_base__swap_u32_argb_abgr(uint32_t u) { uint32_t o = u & 0xFF00FF00; @@ -29,6 +75,8 @@ wuffs_base__swap_u32_argb_abgr(uint32_t u) { return o | (r >> 16) | (b << 16); } +// -------- + static inline uint32_t // wuffs_base__composite_nonpremul_nonpremul_u32_axxx(uint32_t dst_nonpremul, uint32_t src_nonpremul) { @@ -173,6 +221,8 @@ wuffs_base__composite_premul_premul_u32_axxx(uint32_t dst_premul, return (db << 0) | (dg << 8) | (dr << 16) | (da << 24); } +// -------- + wuffs_base__color_u32_argb_premul // wuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* pb, uint32_t x, @@ -214,17 +264,9 @@ wuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* pb, (4 * ((size_t)row[x])))); } - case WUFFS_BASE__PIXEL_FORMAT__BGR_565: { - uint16_t bgr = - wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x))); - uint32_t b5 = 0x1F & (bgr >> 0); - uint32_t b = (b5 << 3) | (b5 >> 2); - uint32_t g6 = 0x3F & (bgr >> 5); - uint32_t g = (g6 << 2) | (g6 >> 4); - uint32_t r5 = 0x1F & (bgr >> 11); - uint32_t r = (r5 << 3) | (r5 >> 2); - return 0xFF000000 | (r << 16) | (g << 8) | (b << 0); - } + case WUFFS_BASE__PIXEL_FORMAT__BGR_565: + return wuffs_base__color_u16_rgb_565__as__color_u32_argb_premul( + wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x)))); case WUFFS_BASE__PIXEL_FORMAT__BGR: return 0xFF000000 | wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x))); @@ -301,15 +343,11 @@ wuffs_base__pixel_buffer__set_color_u32_at( pb->pixcfg.private_impl.pixfmt, color)); break; - case WUFFS_BASE__PIXEL_FORMAT__BGR_565: { - uint32_t b5 = 0x1F & (color >> (8 - 5)); - uint32_t g6 = 0x3F & (color >> (16 - 6)); - uint32_t r5 = 0x1F & (color >> (24 - 5)); - uint32_t bgr565 = (b5 << 0) | (g6 << 5) | (r5 << 11); - wuffs_base__store_u16le__no_bounds_check(row + (2 * ((size_t)x)), - (uint16_t)bgr565); + case WUFFS_BASE__PIXEL_FORMAT__BGR_565: + wuffs_base__store_u16le__no_bounds_check( + row + (2 * ((size_t)x)), + wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(color)); break; - } case WUFFS_BASE__PIXEL_FORMAT__BGR: wuffs_base__store_u24le__no_bounds_check(row + (3 * ((size_t)x)), color); break; @@ -328,7 +366,6 @@ wuffs_base__pixel_buffer__set_color_u32_at( wuffs_base__swap_u32_argb_abgr(color))); break; case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL: - WUFFS_BASE__FALLTHROUGH; case WUFFS_BASE__PIXEL_FORMAT__RGBX: wuffs_base__store_u32le__no_bounds_check( row + (4 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color)); @@ -401,95 +438,11 @@ wuffs_base__pixel_palette__closest_element( } } - return best_index; + return (uint8_t)best_index; } // -------- -static uint64_t // -wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over( - wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { - 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; - uint8_t* s = src.ptr; - size_t n = len; - - // TODO: unroll. - - while (n >= 1) { - uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); - uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); - wuffs_base__store_u32le__no_bounds_check( - d + (0 * 4), - wuffs_base__composite_nonpremul_nonpremul_u32_axxx(d0, s0)); - - s += 1 * 4; - d += 1 * 4; - n -= 1; - } - - return len; -} - -static uint64_t // -wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src( - wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { - 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; - uint8_t* s = src.ptr; - size_t n = len; - - // TODO: unroll. - - while (n >= 1) { - uint32_t s0 = wuffs_base__premul_u32_axxx( - wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))); - wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0); - - s += 1 * 4; - d += 1 * 4; - n -= 1; - } - - return len; -} - -static uint64_t // -wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over( - wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { - 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; - uint8_t* s = src.ptr; - size_t n = len; - - // TODO: unroll. - - while (n >= 1) { - uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); - uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); - wuffs_base__store_u32le__no_bounds_check( - d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0)); - - s += 1 * 4; - d += 1 * 4; - n -= 1; - } - - return len; -} - static uint64_t // wuffs_base__pixel_swizzler__copy_1_1(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -510,10 +463,13 @@ wuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst, return len; } +// -------- + static uint64_t // -wuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { +wuffs_base__pixel_swizzler__bgr_565__index__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { if (dst_palette.len != 1024) { return 0; } @@ -557,6 +513,96 @@ wuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst, return len; } +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + 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; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__store_u32le__no_bounds_check( + d + (0 * 4), + wuffs_base__composite_nonpremul_nonpremul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + 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; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), + wuffs_base__premul_u32_axxx(s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + 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; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__store_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +// -------- + static uint64_t // wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -572,12 +618,13 @@ wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst, const size_t loop_unroll_count = 4; - // The comparison in the while condition is ">", not ">=", because with ">=", - // the last 4-byte store could write past the end of the dst slice. + // The comparison in the while condition is ">", not ">=", because with + // ">=", the last 4-byte store could write past the end of the dst slice. // - // Each 4-byte store writes one too many bytes, but a subsequent store will - // overwrite that with the correct byte. There is always another store, - // whether a 4-byte store in this loop or a 1-byte store in the next loop. + // Each 4-byte store writes one too many bytes, but a subsequent store + // will overwrite that with the correct byte. There is always another + // store, whether a 4-byte store in this loop or a 1-byte store in the + // next loop. while (n > loop_unroll_count) { wuffs_base__store_u32le__no_bounds_check( d + (0 * 3), wuffs_base__load_u32le__no_bounds_check( @@ -600,9 +647,7 @@ wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst, while (n >= 1) { uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr + ((size_t)s[0] * 4)); - d[0] = (uint8_t)(s0 >> 0); - d[1] = (uint8_t)(s0 >> 8); - d[2] = (uint8_t)(s0 >> 16); + wuffs_base__store_u24le__no_bounds_check(d + (0 * 3), s0); s += 1 * 1; d += 1 * 3; @@ -670,6 +715,8 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over( return len; } +// -------- + static uint64_t // wuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -827,52 +874,6 @@ wuffs_base__pixel_swizzler__xxxx__y(wuffs_base__slice_u8 dst, // -------- -static uint64_t // -wuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 src) { - size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; - uint8_t* d = dst.ptr; - uint8_t* s = src.ptr; - - size_t n = len4; - while (n--) { - uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s); - uint32_t b5 = 0x1F & (argb >> (8 - 5)); - uint32_t g6 = 0x3F & (argb >> (16 - 6)); - uint32_t r5 = 0x1F & (argb >> (24 - 5)); - wuffs_base__store_u32le__no_bounds_check( - d, (b5 << 0) | (g6 << 5) | (r5 << 11)); - s += 4; - d += 4; - } - return len4 * 4; -} - -static uint64_t // -wuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 src) { - size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; - uint8_t* d = dst.ptr; - uint8_t* s = src.ptr; - - size_t n = len4; - while (n--) { - uint8_t b0 = s[0]; - uint8_t b1 = s[1]; - uint8_t b2 = s[2]; - uint8_t b3 = s[3]; - d[0] = b2; - d[1] = b1; - d[2] = b0; - d[3] = b3; - s += 4; - d += 4; - } - return len4 * 4; -} - -// -------- - static wuffs_base__pixel_swizzler__func // wuffs_base__pixel_swizzler__prepare__y(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_format dst_format, @@ -930,7 +931,7 @@ wuffs_base__pixel_swizzler__prepare__indexed__bgra_binary( } switch (blend) { case WUFFS_BASE__PIXEL_BLEND__SRC: - return wuffs_base__pixel_swizzler__xx__index__src; + return wuffs_base__pixel_swizzler__bgr_565__index__src; } return NULL; @@ -1005,7 +1006,6 @@ wuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p, break; case WUFFS_BASE__PIXEL_FORMAT__BGR: - case WUFFS_BASE__PIXEL_FORMAT__RGB: // TODO. break; @@ -1015,6 +1015,7 @@ wuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p, case WUFFS_BASE__PIXEL_FORMAT__BGRX: return wuffs_base__pixel_swizzler__xxxx__xxx; + case WUFFS_BASE__PIXEL_FORMAT__RGB: case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY: @@ -1038,12 +1039,10 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul( break; case WUFFS_BASE__PIXEL_FORMAT__BGR: - case WUFFS_BASE__PIXEL_FORMAT__RGB: // TODO. break; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: - case WUFFS_BASE__PIXEL_FORMAT__BGRX: switch (blend) { case WUFFS_BASE__PIXEL_BLEND__SRC: return wuffs_base__pixel_swizzler__copy_4_4; @@ -1062,9 +1061,11 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul( return NULL; case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY: + case WUFFS_BASE__PIXEL_FORMAT__BGRX: // TODO. break; + case WUFFS_BASE__PIXEL_FORMAT__RGB: case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY: diff --git a/internal/cgen/base/image-public.h b/internal/cgen/base/image-public.h index b235f2a1..487f74ae 100644 --- a/internal/cgen/base/image-public.h +++ b/internal/cgen/base/image-public.h @@ -21,6 +21,26 @@ // 0xAARRGGBB (Alpha most significant, Blue least), regardless of endianness. typedef uint32_t wuffs_base__color_u32_argb_premul; +static inline uint16_t // +wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565( + wuffs_base__color_u32_argb_premul c) { + uint32_t r5 = 0xF800 & (c >> 8); + uint32_t g6 = 0x07E0 & (c >> 5); + uint32_t b5 = 0x001F & (c >> 3); + return (uint16_t)(r5 | g6 | b5); +} + +static inline wuffs_base__color_u32_argb_premul // +wuffs_base__color_u16_rgb_565__as__color_u32_argb_premul(uint16_t c) { + uint32_t b5 = 0x1F & (c >> 0); + uint32_t b = (b5 << 3) | (b5 >> 2); + uint32_t g6 = 0x3F & (c >> 5); + uint32_t g = (g6 << 2) | (g6 >> 4); + uint32_t r5 = 0x1F & (c >> 11); + uint32_t r = (r5 << 3) | (r5 >> 2); + return 0xFF000000 | (r << 16) | (g << 8) | (b << 0); +} + static inline uint8_t // wuffs_base__color_u32_argb_premul__as_gray( wuffs_base__color_u32_argb_premul c) { @@ -35,7 +55,8 @@ wuffs_base__color_u32_argb_premul__as_gray( // Note that 19595 + 38470 + 7471 equals 65536, also known as (1 << 16). We // shift by 24, not just by 16, because the return value is 8-bit color, not // 16-bit color. - return ((19595 * cr) + (38470 * cg) + (7471 * cb) + 32768) >> 24; + uint32_t weighted_average = (19595 * cr) + (38470 * cg) + (7471 * cb) + 32768; + return (uint8_t)(weighted_average >> 24); } // wuffs_base__premul_u32_axxx converts from non-premultiplied alpha to diff --git a/internal/cgen/data.go b/internal/cgen/data.go index 6f49c3f9..358f99e2 100644 --- a/internal/cgen/data.go +++ b/internal/cgen/data.go @@ -27,46 +27,55 @@ const baseAllImplC = "" + "" const baseImageImplC = "" + - "// ---------------- Images\n\nconst uint32_t wuffs_base__pixel_format__bits_per_channel[16] = {\n 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,\n 0x08, 0x0A, 0x0C, 0x10, 0x18, 0x20, 0x30, 0x40,\n};\n\nstatic inline uint32_t //\nwuffs_base__swap_u32_argb_abgr(uint32_t u) {\n uint32_t o = u & 0xFF00FF00;\n uint32_t r = u & 0x00FF0000;\n uint32_t b = u & 0x000000FF;\n return o | (r >> 16) | (b << 16);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_nonpremul_nonpremul_u32_axxx(uint32_t dst_nonpremul,\n uint32_t src_nonpremul) {\n // Convert from 8-bit color to 16-bit color.\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_nonpremul >> 24));\n uint32_t dr = 0x101 * (0xFF & (dst_nonpremul >> 16));\n uint32_t dg = 0x101 * (0xFF & (dst_nonpremul >> 8));\n u" + - "int32_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 (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 // Convert from 16-bit color to 8-bit color and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_nonpremul_premul_u32_axxx(uint32_t dst_nonpremul,\n uint32_t src_premul) {\n // Convert from 8-bit color" + - " to 16-bit color.\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 // Con" + - "vert from 16-bit color to 8-bit color and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\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 // Convert from 8-bit color to 16-bit color.\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 and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_premul_premul_u32_axxx(uint32_t dst_premul,\n uint32_t src_premul) {\n // Convert from 8-bit color to 16-bit color.\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 and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nwuffs_base__color_u32_argb_premul //\nwuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* pb,\n uint32_t x,\n uint32_t y) {\n if (!pb || (x >= pb->pixcfg.private_impl.width) ||\n (y >= pb->pixcfg.private_impl.height)) {\n return 0;\n }\n\n if (wuffs_base__pixel_format__is_planar(&pb->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return 0;\n }\n\n size_t stride = pb->private_impl.planes[0].stride;\n uint8_t* row = pb->private_impl.planes[0].ptr + (stride * ((size_t)y));\n\n switch (pb->pixcfg.private_impl.pixfmt.repr) {\n c" + - "ase WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n return wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY: {\n uint8_t* palette = pb->private_impl.planes[3].ptr;\n return wuffs_base__load_u32le__no_bounds_check(palette +\n (4 * ((size_t)row[x])));\n }\n\n // Common formats above. Rarer formats below.\n\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n return 0xFF000000 | (0x00010101 * ((uint32_t)(row[x])));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL: {\n uint8_t* palette = pb->private_impl.planes[3].ptr;\n return wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(palette +\n (4 * ((size_t)row[x]))));\n }\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565: {\n uint16_t bgr =\n " + - " wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x)));\n uint32_t b5 = 0x1F & (bgr >> 0);\n uint32_t b = (b5 << 3) | (b5 >> 2);\n uint32_t g6 = 0x3F & (bgr >> 5);\n uint32_t g = (g6 << 2) | (g6 >> 4);\n uint32_t r5 = 0x1F & (bgr >> 11);\n uint32_t r = (r5 << 3) | (r5 >> 2);\n return 0xFF000000 | (r << 16) | (g << 8) | (b << 0);\n }\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return 0xFF000000 |\n wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x)));\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n return wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n return 0xFF000000 |\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 |\n wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x))))" + - ";\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n return wuffs_base__swap_u32_argb_abgr(wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:\n return wuffs_base__swap_u32_argb_abgr(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 |\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n\n default:\n // TODO: support more formats.\n break;\n }\n\n return 0;\n}\n\nwuffs_base__status //\nwuffs_base__pixel_buffer__set_color_u32_at(\n wuffs_base__pixel_buffer* pb,\n uint32_t x,\n uint32_t y,\n wuffs_base__color_u32_argb_premul color) {\n if (!pb) {\n return wuffs_base__make_status(wuffs_base__error__bad_receiver);\n }\n if ((x >= pb->pixcfg.private_impl.width) ||\n (y >= pb->pixc" + - "fg.private_impl.height)) {\n return wuffs_base__make_status(wuffs_base__error__bad_argument);\n }\n\n if (wuffs_base__pixel_format__is_planar(&pb->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n size_t stride = pb->private_impl.planes[0].stride;\n uint8_t* row = pb->private_impl.planes[0].ptr + (stride * ((size_t)y));\n\n switch (pb->pixcfg.private_impl.pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n wuffs_base__store_u32le__no_bounds_check(row + (4 * ((size_t)x)), color);\n break;\n\n // Common formats above. Rarer formats below.\n\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n wuffs_base__store_u8__no_bounds_check(\n row + ((size_t)x), wuffs_base__color_u32_argb_premul__as_gray(color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY:\n wuffs_base__store_u8__no_bounds_check(\n row + ((size_t)x), wuffs_" + - "base__pixel_palette__closest_element(\n wuffs_base__pixel_buffer__palette(pb),\n pb->pixcfg.private_impl.pixfmt, color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565: {\n uint32_t b5 = 0x1F & (color >> (8 - 5));\n uint32_t g6 = 0x3F & (color >> (16 - 6));\n uint32_t r5 = 0x1F & (color >> (24 - 5));\n uint32_t bgr565 = (b5 << 0) | (g6 << 5) | (r5 << 11);\n wuffs_base__store_u16le__no_bounds_check(row + (2 * ((size_t)x)),\n (uint16_t)bgr565);\n break;\n }\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n wuffs_base__store_u24le__no_bounds_check(row + (3 * ((size_t)x)), color);\n break;\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__nonpremul_u32_axxx(color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n wuffs_base__store_u24le__no_bounds_check(\n r" + - "ow + (3 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__nonpremul_u32_axxx(\n wuffs_base__swap_u32_argb_abgr(color)));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));\n break;\n\n default:\n // TODO: support more formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n return wuffs_base__make_status(NULL);\n}\n\n" + + "// ---------------- Images\n\nconst uint32_t wuffs_base__pixel_format__bits_per_channel[16] = {\n 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,\n 0x08, 0x0A, 0x0C, 0x10, 0x18, 0x20, 0x30, 0x40,\n};\n\n" + + "" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 src) {\n size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n\n size_t n = len4;\n while (n--) {\n uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s);\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 wuffs_base__store_u32le__no_bounds_check(\n d, (r5 << 11) | (g6 << 5) | (b5 << 0));\n s += 4;\n d += 4;\n }\n return len4 * 4;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 src) {\n size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n\n size_t n = len4;\n while (n--) {\n uint8_t b0 = s[0];\n uint8_t b1 = s[1];\n u" + + "int8_t b2 = s[2];\n uint8_t b3 = s[3];\n d[0] = b2;\n d[1] = b1;\n d[2] = b0;\n d[3] = b3;\n s += 4;\n d += 4;\n }\n return len4 * 4;\n}\n\nstatic inline uint32_t //\nwuffs_base__swap_u32_argb_abgr(uint32_t u) {\n uint32_t o = u & 0xFF00FF00;\n uint32_t r = u & 0x00FF0000;\n uint32_t b = u & 0x000000FF;\n return o | (r >> 16) | (b << 16);\n}\n\n" + + "" + + "// --------\n\nstatic inline uint32_t //\nwuffs_base__composite_nonpremul_nonpremul_u32_axxx(uint32_t dst_nonpremul,\n uint32_t src_nonpremul) {\n // Convert from 8-bit color to 16-bit color.\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_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 (nonpremul) over dst (premul).\n da = sa + ((da * ia) / 0xFFFF);\n dr = ((sr * sa) + (dr * i" + + "a)) / 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 // Convert from 16-bit color to 8-bit color and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_nonpremul_premul_u32_axxx(uint32_t dst_nonpremul,\n uint32_t src_premul) {\n // Convert from 8-bit color to 16-bit color.\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 and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\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 // Convert from 8-bit color to 16-bit color.\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 and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_premul_premul_u32_axxx(uint32_t dst_premul,\n " + + " uint32_t src_premul) {\n // Convert from 8-bit color to 16-bit color.\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 and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\n" + + "" + + "// --------\n\nwuffs_base__color_u32_argb_premul //\nwuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* pb,\n uint32_t x,\n uint32_t y) {\n if (!pb || (x >= pb->pixcfg.private_impl.width) ||\n (y >= pb->pixcfg.private_impl.height)) {\n return 0;\n }\n\n if (wuffs_base__pixel_format__is_planar(&pb->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return 0;\n }\n\n size_t stride = pb->private_impl.planes[0].stride;\n uint8_t* row = pb->private_impl.planes[0].ptr + (stride * ((size_t)y));\n\n switch (pb->pixcfg.private_impl.pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n return wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY: {\n uint8_t* palette = pb->private_impl.planes[3].ptr;\n return wu" + + "ffs_base__load_u32le__no_bounds_check(palette +\n (4 * ((size_t)row[x])));\n }\n\n // Common formats above. Rarer formats below.\n\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n return 0xFF000000 | (0x00010101 * ((uint32_t)(row[x])));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL: {\n uint8_t* palette = pb->private_impl.planes[3].ptr;\n return wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(palette +\n (4 * ((size_t)row[x]))));\n }\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n return wuffs_base__color_u16_rgb_565__as__color_u32_argb_premul(\n wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return 0xFF000000 |\n wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x)));\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n return wuffs_base__premul_u32_axxx(\n " + + " wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n return 0xFF000000 |\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 |\n wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n return wuffs_base__swap_u32_argb_abgr(wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:\n return wuffs_base__swap_u32_argb_abgr(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 |\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n\n defaul" + + "t:\n // TODO: support more formats.\n break;\n }\n\n return 0;\n}\n\nwuffs_base__status //\nwuffs_base__pixel_buffer__set_color_u32_at(\n wuffs_base__pixel_buffer* pb,\n uint32_t x,\n uint32_t y,\n wuffs_base__color_u32_argb_premul color) {\n if (!pb) {\n return wuffs_base__make_status(wuffs_base__error__bad_receiver);\n }\n if ((x >= pb->pixcfg.private_impl.width) ||\n (y >= pb->pixcfg.private_impl.height)) {\n return wuffs_base__make_status(wuffs_base__error__bad_argument);\n }\n\n if (wuffs_base__pixel_format__is_planar(&pb->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n size_t stride = pb->private_impl.planes[0].stride;\n uint8_t* row = pb->private_impl.planes[0].ptr + (stride * ((size_t)y));\n\n switch (pb->pixcfg.private_impl.pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n wuffs_base__store_u32le__no_bounds_check(row + (4 *" + + " ((size_t)x)), color);\n break;\n\n // Common formats above. Rarer formats below.\n\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n wuffs_base__store_u8__no_bounds_check(\n row + ((size_t)x), wuffs_base__color_u32_argb_premul__as_gray(color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY:\n wuffs_base__store_u8__no_bounds_check(\n row + ((size_t)x), wuffs_base__pixel_palette__closest_element(\n wuffs_base__pixel_buffer__palette(pb),\n pb->pixcfg.private_impl.pixfmt, color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n wuffs_base__store_u16le__no_bounds_check(\n row + (2 * ((size_t)x)),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(color));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n wuffs_base__store_u24le__no_bounds_check(row + (3 * ((size_t)x)), color);\n break;\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n wuffs_base__sto" + + "re_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__nonpremul_u32_axxx(color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n wuffs_base__store_u24le__no_bounds_check(\n row + (3 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__nonpremul_u32_axxx(\n wuffs_base__swap_u32_argb_abgr(color)));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));\n break;\n\n default:\n // TODO: support more formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n return wuffs_base__make_status(NULL);\n}\n\n" + "" + "// --------\n\nuint8_t //\nwuffs_base__pixel_palette__closest_element(\n wuffs_base__slice_u8 palette_slice,\n wuffs_base__pixel_format palette_format,\n wuffs_base__color_u32_argb_premul c) {\n size_t n = palette_slice.len / 4;\n if (n > 256) {\n n = 256;\n }\n size_t best_index = 0;\n uint64_t best_score = 0xFFFFFFFFFFFFFFFF;\n\n // Work in 16-bit color.\n uint32_t ca = 0x101 * (0xFF & (c >> 24));\n uint32_t cr = 0x101 * (0xFF & (c >> 16));\n uint32_t cg = 0x101 * (0xFF & (c >> 8));\n uint32_t cb = 0x101 * (0xFF & (c >> 0));\n\n switch (palette_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY: {\n bool nonpremul = palette_format.repr ==\n WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL;\n\n size_t i;\n for (i = 0; i < n; i++) {\n // Work in 16-bit color.\n uint32_t pb = 0x101 * ((uint32_t)(palette_slice.ptr[(4 * i) + 0]))" + - ";\n uint32_t pg = 0x101 * ((uint32_t)(palette_slice.ptr[(4 * i) + 1]));\n uint32_t pr = 0x101 * ((uint32_t)(palette_slice.ptr[(4 * i) + 2]));\n uint32_t pa = 0x101 * ((uint32_t)(palette_slice.ptr[(4 * i) + 3]));\n\n // Convert to premultiplied alpha.\n if (nonpremul && (pa != 0xFFFF)) {\n pb = (pb * pa) / 0xFFFF;\n pg = (pg * pa) / 0xFFFF;\n pr = (pr * pa) / 0xFFFF;\n }\n\n // These deltas are conceptually int32_t (signed) but after squaring,\n // it's equivalent to work in uint32_t (unsigned).\n pb -= cb;\n pg -= cg;\n pr -= cr;\n pa -= ca;\n uint64_t score = ((uint64_t)(pb * pb)) + ((uint64_t)(pg * pg)) +\n ((uint64_t)(pr * pr)) + ((uint64_t)(pa * pa));\n if (best_score > score) {\n best_score = score;\n best_index = i;\n }\n }\n break;\n }\n }\n\n return best_index;\n}\n\n" + + ";\n uint32_t pg = 0x101 * ((uint32_t)(palette_slice.ptr[(4 * i) + 1]));\n uint32_t pr = 0x101 * ((uint32_t)(palette_slice.ptr[(4 * i) + 2]));\n uint32_t pa = 0x101 * ((uint32_t)(palette_slice.ptr[(4 * i) + 3]));\n\n // Convert to premultiplied alpha.\n if (nonpremul && (pa != 0xFFFF)) {\n pb = (pb * pa) / 0xFFFF;\n pg = (pg * pa) / 0xFFFF;\n pr = (pr * pa) / 0xFFFF;\n }\n\n // These deltas are conceptually int32_t (signed) but after squaring,\n // it's equivalent to work in uint32_t (unsigned).\n pb -= cb;\n pg -= cg;\n pr -= cr;\n pa -= ca;\n uint64_t score = ((uint64_t)(pb * pb)) + ((uint64_t)(pg * pg)) +\n ((uint64_t)(pr * pr)) + ((uint64_t)(pa * pa));\n if (best_score > score) {\n best_score = score;\n best_index = i;\n }\n }\n break;\n }\n }\n\n return (uint8_t)best_index;\n}\n\n" + "" + - "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__load_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__store_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_premul__bgra_nonpremul__src(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), 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__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__load_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__store_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__copy_1_1(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n return wuffs_base__slice_u8__copy_from_slice(dst, src);\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 if (len > 0) {\n memmove(dst.ptr, src.ptr, len * 4);\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n " + - " wuffs_base__slice_u8 src) {\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 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__store_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (1 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (2 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (3 * 2), wuffs_base__load_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__store_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__load_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__xxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst.len / 3;\n size_t len = dst_len3 < src.len ? dst_len3 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n // The comparison in the while condition is \">\", not \">=\", because with \">=\",\n // the last 4-byte store could write past the end of the dst slice.\n //\n // Each 4-byte store wri" + - "tes one too many bytes, but a subsequent store will\n // overwrite that with the correct byte. There is always another store,\n // whether a 4-byte store in this loop or a 1-byte store in the next loop.\n while (n > loop_unroll_count) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (1 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (2 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (3 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 3;\n n -= loop_unroll_count;\n " + - "}\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n d[0] = (uint8_t)(s0 >> 0);\n d[1] = (uint8_t)(s0 >> 8);\n d[2] = (uint8_t)(s0 >> 16);\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst.len / 3;\n size_t len = dst_len3 < src.len ? dst_len3 : src.len;\n uint8_t* d = dst.ptr;\n 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 uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u24le__no_bounds_check(d + (0" + - " * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u24le__no_bounds_check(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u24le__no_bounds_check(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[3] * 4));\n if (s3) {\n wuffs_base__store_u24le__no_bounds_check(d + (3 * 4), s3);\n }\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 3;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n" + - " wuffs_base__store_u24le__no_bounds_check(d + (0 * 4), s0);\n }\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__store_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (1 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__" + - "store_u32le__no_bounds_check(\n d + (2 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (3 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 4;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\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__xxxx__index_binary_alpha__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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 uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u32le__no_bounds_check(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u32le__no_bounds_check(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[3] * 4));\n if (s3) {\n w" + - "uffs_base__store_u32le__no_bounds_check(d + (3 * 4), s3);\n }\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 4;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);\n }\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__xxxx__xxx(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4),\n 0xFF00000" + - "0 | wuffs_base__load_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__xxxx__y(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4), 0xFF000000 | (0x010101 * (uint32_t)s[0]));\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_1_1(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n return wuffs_base__slice_u8__copy_from_slice(dst, src);\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 if (len > 0) {\n memmove(dst.ptr, src.ptr, len * 4);\n }\n return len;\n}\n\n" + "" + - "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 src) {\n size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n\n size_t n = len4;\n while (n--) {\n uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s);\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 wuffs_base__store_u32le__no_bounds_check(\n d, (b5 << 0) | (g6 << 5) | (r5 << 11));\n s += 4;\n d += 4;\n }\n return len4 * 4;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 src) {\n size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n\n size_t n = len4;\n while (n--) {\n uint8_t b0 = s[0];\n uint8_t b1 = s[1];\n u" + - "int8_t b2 = s[2];\n uint8_t b3 = s[3];\n d[0] = b2;\n d[1] = b1;\n d[2] = b0;\n d[3] = b3;\n s += 4;\n d += 4;\n }\n return len4 * 4;\n}\n\n" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__index__src(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__store_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (1 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (2 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u16le__no_bounds_che" + + "ck(\n d + (3 * 2), wuffs_base__load_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__store_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__load_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\n" + + "" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__load_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__store_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\n" + + "" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__load_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4),\n wuffs_base__premul_u32_axxx(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__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__load_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__store_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\n" + + "" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst.len / 3;\n size_t len = dst_len3 < src.len ? dst_len3 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n // The comparison in the while condition is \">\", not \">=\", because with\n // \">=\", the last 4-byte store could write past the end of the dst slice.\n //\n // Each 4-byte store writes one too many bytes, but a subsequent store\n // will overwrite that with the correct byte. There is always another\n // store, whether a 4-byte store in this loop or a 1-byte store in the\n // next loop.\n while (n > loop_unroll_count) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 3), wuffs_base__load_u32le__no_bounds_c" + + "heck(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (1 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (2 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (3 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 3;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n wuffs_base__store_u24le__no_bounds_check(d + (0 * 3), s0);\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__" + + "pixel_swizzler__xxx__index_binary_alpha__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst.len / 3;\n size_t len = dst_len3 < src.len ? dst_len3 : src.len;\n uint8_t* d = dst.ptr;\n 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 uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u24le__no_bounds_check(d + (0 * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u24le__no_bounds_check(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n " + + " ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u24le__no_bounds_check(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[3] * 4));\n if (s3) {\n wuffs_base__store_u24le__no_bounds_check(d + (3 * 4), s3);\n }\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 3;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u24le__no_bounds_check(d + (0 * 4), s0);\n }\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\n" + + "" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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__store_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (1 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (2 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_pale" + + "tte.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (3 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 4;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\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__xxxx__index_binary_alpha__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 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 uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u32le__no_bounds_check(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u32le__no_bounds_check(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[3] * 4));\n if (s3) {\n wuffs_base__store_u32le__no_bounds_check(d + (3 * 4), s3);\n }\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count *" + + " 4;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);\n }\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__xxxx__xxx(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4),\n 0xFF000000 | wuffs_base__load_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__xxxx__y(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4), 0xFF000000 | (0x010101 * (uint32_t)s[0]));\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" + "" + "// --------\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__y(wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_format,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n // TODO.\n break;\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 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 wuf" + "fs_base__pixel_swizzler__xxxx__y;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__indexed__bgra_binary(\n wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_format,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY:\n if (wuffs_base__slice_u8__copy_from_slice(dst_palette, src_palette) !=\n 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__copy_1_1;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n if (wuffs_base__pixel_swizzler__squash_bgr_565_888(dst_palette,\n src_palette) != 1024) {\n " + - " return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xx__index__src;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n if (wuffs_base__slice_u8__copy_from_slice(dst_palette, src_palette) !=\n 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxx__index__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over;\n }\n return NULL;\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 if (wuffs_base__slice_u8__copy_from_slice(dst_palette, src_palette) !=\n 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxxx__index__sr" + - "c;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxxx__index_binary_alpha__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n if (wuffs_base__pixel_swizzler__swap_rgbx_bgrx(dst_palette,\n src_palette) != 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxx__index__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over;\n }\n return NULL;\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 if (wuffs_base__pixel_swizzler__swap_rgbx_bgrx(dst_palette,\n src_palette) != 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__P" + - "IXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxxx__index__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxxx__index_binary_alpha__src_over;\n }\n return NULL;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_format,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGR" + - "A_BINARY:\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n return wuffs_base__pixel_swizzler__xxxx__xxx;\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 // TODO.\n break;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgra_nonpremul(\n wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_format,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\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__BGRA_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__BGRA_BINARY:\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 // TODO.\n break;\n }\n return NULL;\n}\n\n" + + " return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgr_565__index__src;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n if (wuffs_base__slice_u8__copy_from_slice(dst_palette, src_palette) !=\n 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxx__index__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over;\n }\n return NULL;\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 if (wuffs_base__slice_u8__copy_from_slice(dst_palette, src_palette) !=\n 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxxx__inde" + + "x__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxxx__index_binary_alpha__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n if (wuffs_base__pixel_swizzler__swap_rgbx_bgrx(dst_palette,\n src_palette) != 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxx__index__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over;\n }\n return NULL;\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 if (wuffs_base__pixel_swizzler__swap_rgbx_bgrx(dst_palette,\n src_palette) != 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BA" + + "SE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__xxxx__index__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__xxxx__index_binary_alpha__src_over;\n }\n return NULL;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_format,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n // TODO.\n break;\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__PIXE" + + "L_FORMAT__BGRX:\n return wuffs_base__pixel_swizzler__xxxx__xxx;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\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 // TODO.\n break;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgra_nonpremul(\n wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_format,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n // TODO.\n break;\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__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__BGRA_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__BGRA_BINARY:\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\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 // TODO.\n break;\n }\n return NULL;\n}\n\n" + "" + "// --------\n\nwuffs_base__status //\nwuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_format,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__pixel_format src_format,\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\n // TODO: support many more formats.\n\n wuffs_base__pixel_swizzler__func func = NULL;\n\n switch (src_format.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n func = wuffs_base__pixel_swizzler__prepare__y(p, dst_format, dst_palette,\n 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_format, 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_format, 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_format, dst_palette, src_palette, blend);\n break;\n }\n\n p->private_impl.func = func;\n return wuffs_base__make_status(\n func ? NULL : wuffs_base__error__unsupported_pixel_swizzler_option);\n}\n\nuint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved(\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, dst_palette, src);\n }\n return 0;\n}\n" + @@ -235,9 +244,10 @@ const baseImagePrivateH = "" + "" const baseImagePublicH = "" + - "// ---------------- Images\n\n// wuffs_base__color_u32_argb_premul is an 8 bit per channel premultiplied\n// Alpha, Red, Green, Blue color, as a uint32_t value. Its value is always\n// 0xAARRGGBB (Alpha most significant, Blue least), regardless of endianness.\ntypedef uint32_t wuffs_base__color_u32_argb_premul;\n\nstatic inline uint8_t //\nwuffs_base__color_u32_argb_premul__as_gray(\n wuffs_base__color_u32_argb_premul c) {\n // Work in 16-bit color.\n uint32_t cr = 0x101 * (0xFF & (c >> 16));\n uint32_t cg = 0x101 * (0xFF & (c >> 8));\n uint32_t cb = 0x101 * (0xFF & (c >> 0));\n\n // These coefficients (the fractions 0.299, 0.587 and 0.114) are the same\n // as those given by the JFIF specification.\n //\n // Note that 19595 + 38470 + 7471 equals 65536, also known as (1 << 16). We\n // shift by 24, not just by 16, because the return value is 8-bit color, not\n // 16-bit color.\n return ((19595 * cr) + (38470 * cg) + (7471 * cb) + 32768) >> 24;\n}\n\n// wuffs_base__premul_u32_axxx converts from non-premultiplied alpha " + - "to\n// premultiplied alpha. The \"axxx\" means either \"argb\" or \"abgr\".\nstatic inline uint32_t //\nwuffs_base__premul_u32_axxx(uint32_t nonpremul) {\n // Multiplying by 0x101 (twice, once for alpha and once for color) converts\n // from 8-bit to 16-bit color. Shifting right by 8 undoes that.\n //\n // Working in the higher bit depth can produce slightly different (and\n // arguably slightly more accurate) results. For example, given 8-bit blue\n // and alpha of 0x80 and 0x81:\n //\n // - ((0x80 * 0x81 ) / 0xFF ) = 0x40 = 0x40\n // - ((0x8080 * 0x8181) / 0xFFFF) >> 8 = 0x4101 >> 8 = 0x41\n uint32_t a = 0xFF & (nonpremul >> 24);\n uint32_t a16 = a * (0x101 * 0x101);\n\n uint32_t r = 0xFF & (nonpremul >> 16);\n r = ((r * a16) / 0xFFFF) >> 8;\n uint32_t g = 0xFF & (nonpremul >> 8);\n g = ((g * a16) / 0xFFFF) >> 8;\n uint32_t b = 0xFF & (nonpremul >> 0);\n b = ((b * a16) / 0xFFFF) >> 8;\n\n return (a << 24) | (r << 16) | (g << 8) | (b << 0);\n}\n\n// wuffs_base__nonpremul_u32_axxx converts from premultip" + - "lied alpha to\n// non-premultiplied alpha. The \"axxx\" means either \"argb\" or \"abgr\".\nstatic inline uint32_t //\nwuffs_base__nonpremul_u32_axxx(uint32_t premul) {\n uint32_t a = 0xFF & (premul >> 24);\n if (a == 0xFF) {\n return premul;\n } else if (a == 0) {\n return 0;\n }\n uint32_t a16 = a * 0x101;\n\n uint32_t r = 0xFF & (premul >> 16);\n r = ((r * (0x101 * 0xFFFF)) / a16) >> 8;\n uint32_t g = 0xFF & (premul >> 8);\n g = ((g * (0x101 * 0xFFFF)) / a16) >> 8;\n uint32_t b = 0xFF & (premul >> 0);\n b = ((b * (0x101 * 0xFFFF)) / a16) >> 8;\n\n return (a << 24) | (r << 16) | (g << 8) | (b << 0);\n}\n\n" + + "// ---------------- Images\n\n// wuffs_base__color_u32_argb_premul is an 8 bit per channel premultiplied\n// Alpha, Red, Green, Blue color, as a uint32_t value. Its value is always\n// 0xAARRGGBB (Alpha most significant, Blue least), regardless of endianness.\ntypedef uint32_t wuffs_base__color_u32_argb_premul;\n\nstatic inline uint16_t //\nwuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__color_u32_argb_premul c) {\n uint32_t r5 = 0xF800 & (c >> 8);\n uint32_t g6 = 0x07E0 & (c >> 5);\n uint32_t b5 = 0x001F & (c >> 3);\n return (uint16_t)(r5 | g6 | b5);\n}\n\nstatic inline wuffs_base__color_u32_argb_premul //\nwuffs_base__color_u16_rgb_565__as__color_u32_argb_premul(uint16_t c) {\n uint32_t b5 = 0x1F & (c >> 0);\n uint32_t b = (b5 << 3) | (b5 >> 2);\n uint32_t g6 = 0x3F & (c >> 5);\n uint32_t g = (g6 << 2) | (g6 >> 4);\n uint32_t r5 = 0x1F & (c >> 11);\n uint32_t r = (r5 << 3) | (r5 >> 2);\n return 0xFF000000 | (r << 16) | (g << 8) | (b << 0);\n}\n\nstatic inline uint8_t //\nwuffs_base__color_u32_a" + + "rgb_premul__as_gray(\n wuffs_base__color_u32_argb_premul c) {\n // Work in 16-bit color.\n uint32_t cr = 0x101 * (0xFF & (c >> 16));\n uint32_t cg = 0x101 * (0xFF & (c >> 8));\n uint32_t cb = 0x101 * (0xFF & (c >> 0));\n\n // These coefficients (the fractions 0.299, 0.587 and 0.114) are the same\n // as those given by the JFIF specification.\n //\n // Note that 19595 + 38470 + 7471 equals 65536, also known as (1 << 16). We\n // shift by 24, not just by 16, because the return value is 8-bit color, not\n // 16-bit color.\n uint32_t weighted_average = (19595 * cr) + (38470 * cg) + (7471 * cb) + 32768;\n return (uint8_t)(weighted_average >> 24);\n}\n\n// wuffs_base__premul_u32_axxx converts from non-premultiplied alpha to\n// premultiplied alpha. The \"axxx\" means either \"argb\" or \"abgr\".\nstatic inline uint32_t //\nwuffs_base__premul_u32_axxx(uint32_t nonpremul) {\n // Multiplying by 0x101 (twice, once for alpha and once for color) converts\n // from 8-bit to 16-bit color. Shifting right by 8 undoes that.\n //\n // W" + + "orking in the higher bit depth can produce slightly different (and\n // arguably slightly more accurate) results. For example, given 8-bit blue\n // and alpha of 0x80 and 0x81:\n //\n // - ((0x80 * 0x81 ) / 0xFF ) = 0x40 = 0x40\n // - ((0x8080 * 0x8181) / 0xFFFF) >> 8 = 0x4101 >> 8 = 0x41\n uint32_t a = 0xFF & (nonpremul >> 24);\n uint32_t a16 = a * (0x101 * 0x101);\n\n uint32_t r = 0xFF & (nonpremul >> 16);\n r = ((r * a16) / 0xFFFF) >> 8;\n uint32_t g = 0xFF & (nonpremul >> 8);\n g = ((g * a16) / 0xFFFF) >> 8;\n uint32_t b = 0xFF & (nonpremul >> 0);\n b = ((b * a16) / 0xFFFF) >> 8;\n\n return (a << 24) | (r << 16) | (g << 8) | (b << 0);\n}\n\n// wuffs_base__nonpremul_u32_axxx converts from premultiplied alpha to\n// non-premultiplied alpha. The \"axxx\" means either \"argb\" or \"abgr\".\nstatic inline uint32_t //\nwuffs_base__nonpremul_u32_axxx(uint32_t premul) {\n uint32_t a = 0xFF & (premul >> 24);\n if (a == 0xFF) {\n return premul;\n } else if (a == 0) {\n return 0;\n }\n uint32_t a16 = a * 0x" + + "101;\n\n uint32_t r = 0xFF & (premul >> 16);\n r = ((r * (0x101 * 0xFFFF)) / a16) >> 8;\n uint32_t g = 0xFF & (premul >> 8);\n g = ((g * (0x101 * 0xFFFF)) / a16) >> 8;\n uint32_t b = 0xFF & (premul >> 0);\n b = ((b * (0x101 * 0xFFFF)) / a16) >> 8;\n\n return (a << 24) | (r << 16) | (g << 8) | (b << 0);\n}\n\n" + "" + "// --------\n\ntypedef uint8_t wuffs_base__pixel_blend;\n\n// wuffs_base__pixel_blend encodes how to blend source and destination pixels,\n// accounting for transparency. It encompasses the Porter-Duff compositing\n// operators as well as the other blending modes defined by PDF.\n//\n// TODO: implement the other modes.\n#define WUFFS_BASE__PIXEL_BLEND__SRC ((wuffs_base__pixel_blend)0)\n#define WUFFS_BASE__PIXEL_BLEND__SRC_OVER ((wuffs_base__pixel_blend)1)\n\n" + "" + diff --git a/release/c/wuffs-unsupported-snapshot.c b/release/c/wuffs-unsupported-snapshot.c index 876c908f..5e0fe7e1 100644 --- a/release/c/wuffs-unsupported-snapshot.c +++ b/release/c/wuffs-unsupported-snapshot.c @@ -2507,6 +2507,26 @@ wuffs_base__malloc_slice_u64(void* (*malloc_func)(size_t), uint64_t num_u64) { // 0xAARRGGBB (Alpha most significant, Blue least), regardless of endianness. typedef uint32_t wuffs_base__color_u32_argb_premul; +static inline uint16_t // +wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565( + wuffs_base__color_u32_argb_premul c) { + uint32_t r5 = 0xF800 & (c >> 8); + uint32_t g6 = 0x07E0 & (c >> 5); + uint32_t b5 = 0x001F & (c >> 3); + return (uint16_t)(r5 | g6 | b5); +} + +static inline wuffs_base__color_u32_argb_premul // +wuffs_base__color_u16_rgb_565__as__color_u32_argb_premul(uint16_t c) { + uint32_t b5 = 0x1F & (c >> 0); + uint32_t b = (b5 << 3) | (b5 >> 2); + uint32_t g6 = 0x3F & (c >> 5); + uint32_t g = (g6 << 2) | (g6 >> 4); + uint32_t r5 = 0x1F & (c >> 11); + uint32_t r = (r5 << 3) | (r5 >> 2); + return 0xFF000000 | (r << 16) | (g << 8) | (b << 0); +} + static inline uint8_t // wuffs_base__color_u32_argb_premul__as_gray( wuffs_base__color_u32_argb_premul c) { @@ -2521,7 +2541,8 @@ wuffs_base__color_u32_argb_premul__as_gray( // Note that 19595 + 38470 + 7471 equals 65536, also known as (1 << 16). We // shift by 24, not just by 16, because the return value is 8-bit color, not // 16-bit color. - return ((19595 * cr) + (38470 * cg) + (7471 * cb) + 32768) >> 24; + uint32_t weighted_average = (19595 * cr) + (38470 * cg) + (7471 * cb) + 32768; + return (uint8_t)(weighted_average >> 24); } // wuffs_base__premul_u32_axxx converts from non-premultiplied alpha to @@ -8506,6 +8527,52 @@ const uint32_t wuffs_base__pixel_format__bits_per_channel[16] = { 0x08, 0x0A, 0x0C, 0x10, 0x18, 0x20, 0x30, 0x40, }; +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 src) { + size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + + size_t n = len4; + while (n--) { + uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s); + uint32_t b5 = 0x1F & (argb >> (8 - 5)); + uint32_t g6 = 0x3F & (argb >> (16 - 6)); + uint32_t r5 = 0x1F & (argb >> (24 - 5)); + wuffs_base__store_u32le__no_bounds_check( + d, (r5 << 11) | (g6 << 5) | (b5 << 0)); + s += 4; + d += 4; + } + return len4 * 4; +} + +static uint64_t // +wuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 src) { + size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + + size_t n = len4; + while (n--) { + uint8_t b0 = s[0]; + uint8_t b1 = s[1]; + uint8_t b2 = s[2]; + uint8_t b3 = s[3]; + d[0] = b2; + d[1] = b1; + d[2] = b0; + d[3] = b3; + s += 4; + d += 4; + } + return len4 * 4; +} + static inline uint32_t // wuffs_base__swap_u32_argb_abgr(uint32_t u) { uint32_t o = u & 0xFF00FF00; @@ -8514,6 +8581,8 @@ wuffs_base__swap_u32_argb_abgr(uint32_t u) { return o | (r >> 16) | (b << 16); } +// -------- + static inline uint32_t // wuffs_base__composite_nonpremul_nonpremul_u32_axxx(uint32_t dst_nonpremul, uint32_t src_nonpremul) { @@ -8658,6 +8727,8 @@ wuffs_base__composite_premul_premul_u32_axxx(uint32_t dst_premul, return (db << 0) | (dg << 8) | (dr << 16) | (da << 24); } +// -------- + wuffs_base__color_u32_argb_premul // wuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* pb, uint32_t x, @@ -8699,17 +8770,9 @@ wuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* pb, (4 * ((size_t)row[x])))); } - case WUFFS_BASE__PIXEL_FORMAT__BGR_565: { - uint16_t bgr = - wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x))); - uint32_t b5 = 0x1F & (bgr >> 0); - uint32_t b = (b5 << 3) | (b5 >> 2); - uint32_t g6 = 0x3F & (bgr >> 5); - uint32_t g = (g6 << 2) | (g6 >> 4); - uint32_t r5 = 0x1F & (bgr >> 11); - uint32_t r = (r5 << 3) | (r5 >> 2); - return 0xFF000000 | (r << 16) | (g << 8) | (b << 0); - } + case WUFFS_BASE__PIXEL_FORMAT__BGR_565: + return wuffs_base__color_u16_rgb_565__as__color_u32_argb_premul( + wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x)))); case WUFFS_BASE__PIXEL_FORMAT__BGR: return 0xFF000000 | wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x))); @@ -8786,15 +8849,11 @@ wuffs_base__pixel_buffer__set_color_u32_at( pb->pixcfg.private_impl.pixfmt, color)); break; - case WUFFS_BASE__PIXEL_FORMAT__BGR_565: { - uint32_t b5 = 0x1F & (color >> (8 - 5)); - uint32_t g6 = 0x3F & (color >> (16 - 6)); - uint32_t r5 = 0x1F & (color >> (24 - 5)); - uint32_t bgr565 = (b5 << 0) | (g6 << 5) | (r5 << 11); - wuffs_base__store_u16le__no_bounds_check(row + (2 * ((size_t)x)), - (uint16_t)bgr565); + case WUFFS_BASE__PIXEL_FORMAT__BGR_565: + wuffs_base__store_u16le__no_bounds_check( + row + (2 * ((size_t)x)), + wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(color)); break; - } case WUFFS_BASE__PIXEL_FORMAT__BGR: wuffs_base__store_u24le__no_bounds_check(row + (3 * ((size_t)x)), color); break; @@ -8813,7 +8872,6 @@ wuffs_base__pixel_buffer__set_color_u32_at( wuffs_base__swap_u32_argb_abgr(color))); break; case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL: - WUFFS_BASE__FALLTHROUGH; case WUFFS_BASE__PIXEL_FORMAT__RGBX: wuffs_base__store_u32le__no_bounds_check( row + (4 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color)); @@ -8886,95 +8944,11 @@ wuffs_base__pixel_palette__closest_element( } } - return best_index; + return (uint8_t)best_index; } // -------- -static uint64_t // -wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over( - wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { - 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; - uint8_t* s = src.ptr; - size_t n = len; - - // TODO: unroll. - - while (n >= 1) { - uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); - uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); - wuffs_base__store_u32le__no_bounds_check( - d + (0 * 4), - wuffs_base__composite_nonpremul_nonpremul_u32_axxx(d0, s0)); - - s += 1 * 4; - d += 1 * 4; - n -= 1; - } - - return len; -} - -static uint64_t // -wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src( - wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { - 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; - uint8_t* s = src.ptr; - size_t n = len; - - // TODO: unroll. - - while (n >= 1) { - uint32_t s0 = wuffs_base__premul_u32_axxx( - wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))); - wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0); - - s += 1 * 4; - d += 1 * 4; - n -= 1; - } - - return len; -} - -static uint64_t // -wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over( - wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { - 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; - uint8_t* s = src.ptr; - size_t n = len; - - // TODO: unroll. - - while (n >= 1) { - uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); - uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); - wuffs_base__store_u32le__no_bounds_check( - d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0)); - - s += 1 * 4; - d += 1 * 4; - n -= 1; - } - - return len; -} - static uint64_t // wuffs_base__pixel_swizzler__copy_1_1(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -8995,10 +8969,13 @@ wuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst, return len; } +// -------- + static uint64_t // -wuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 dst_palette, - wuffs_base__slice_u8 src) { +wuffs_base__pixel_swizzler__bgr_565__index__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { if (dst_palette.len != 1024) { return 0; } @@ -9042,6 +9019,96 @@ wuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst, return len; } +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + 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; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__store_u32le__no_bounds_check( + d + (0 * 4), + wuffs_base__composite_nonpremul_nonpremul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + 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; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), + wuffs_base__premul_u32_axxx(s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + 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; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__store_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +// -------- + static uint64_t // wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -9057,12 +9124,13 @@ wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst, const size_t loop_unroll_count = 4; - // The comparison in the while condition is ">", not ">=", because with ">=", - // the last 4-byte store could write past the end of the dst slice. + // The comparison in the while condition is ">", not ">=", because with + // ">=", the last 4-byte store could write past the end of the dst slice. // - // Each 4-byte store writes one too many bytes, but a subsequent store will - // overwrite that with the correct byte. There is always another store, - // whether a 4-byte store in this loop or a 1-byte store in the next loop. + // Each 4-byte store writes one too many bytes, but a subsequent store + // will overwrite that with the correct byte. There is always another + // store, whether a 4-byte store in this loop or a 1-byte store in the + // next loop. while (n > loop_unroll_count) { wuffs_base__store_u32le__no_bounds_check( d + (0 * 3), wuffs_base__load_u32le__no_bounds_check( @@ -9085,9 +9153,7 @@ wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst, while (n >= 1) { uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr + ((size_t)s[0] * 4)); - d[0] = (uint8_t)(s0 >> 0); - d[1] = (uint8_t)(s0 >> 8); - d[2] = (uint8_t)(s0 >> 16); + wuffs_base__store_u24le__no_bounds_check(d + (0 * 3), s0); s += 1 * 1; d += 1 * 3; @@ -9155,6 +9221,8 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over( return len; } +// -------- + static uint64_t // wuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -9312,52 +9380,6 @@ wuffs_base__pixel_swizzler__xxxx__y(wuffs_base__slice_u8 dst, // -------- -static uint64_t // -wuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 src) { - size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; - uint8_t* d = dst.ptr; - uint8_t* s = src.ptr; - - size_t n = len4; - while (n--) { - uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s); - uint32_t b5 = 0x1F & (argb >> (8 - 5)); - uint32_t g6 = 0x3F & (argb >> (16 - 6)); - uint32_t r5 = 0x1F & (argb >> (24 - 5)); - wuffs_base__store_u32le__no_bounds_check( - d, (b5 << 0) | (g6 << 5) | (r5 << 11)); - s += 4; - d += 4; - } - return len4 * 4; -} - -static uint64_t // -wuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst, - wuffs_base__slice_u8 src) { - size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4; - uint8_t* d = dst.ptr; - uint8_t* s = src.ptr; - - size_t n = len4; - while (n--) { - uint8_t b0 = s[0]; - uint8_t b1 = s[1]; - uint8_t b2 = s[2]; - uint8_t b3 = s[3]; - d[0] = b2; - d[1] = b1; - d[2] = b0; - d[3] = b3; - s += 4; - d += 4; - } - return len4 * 4; -} - -// -------- - static wuffs_base__pixel_swizzler__func // wuffs_base__pixel_swizzler__prepare__y(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_format dst_format, @@ -9415,7 +9437,7 @@ wuffs_base__pixel_swizzler__prepare__indexed__bgra_binary( } switch (blend) { case WUFFS_BASE__PIXEL_BLEND__SRC: - return wuffs_base__pixel_swizzler__xx__index__src; + return wuffs_base__pixel_swizzler__bgr_565__index__src; } return NULL; @@ -9490,7 +9512,6 @@ wuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p, break; case WUFFS_BASE__PIXEL_FORMAT__BGR: - case WUFFS_BASE__PIXEL_FORMAT__RGB: // TODO. break; @@ -9500,6 +9521,7 @@ wuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p, case WUFFS_BASE__PIXEL_FORMAT__BGRX: return wuffs_base__pixel_swizzler__xxxx__xxx; + case WUFFS_BASE__PIXEL_FORMAT__RGB: case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY: @@ -9523,12 +9545,10 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul( break; case WUFFS_BASE__PIXEL_FORMAT__BGR: - case WUFFS_BASE__PIXEL_FORMAT__RGB: // TODO. break; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: - case WUFFS_BASE__PIXEL_FORMAT__BGRX: switch (blend) { case WUFFS_BASE__PIXEL_BLEND__SRC: return wuffs_base__pixel_swizzler__copy_4_4; @@ -9547,9 +9567,11 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul( return NULL; case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY: + case WUFFS_BASE__PIXEL_FORMAT__BGRX: // TODO. break; + case WUFFS_BASE__PIXEL_FORMAT__RGB: case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL: case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY: