diff --git a/internal/cgen/base/image-impl.c b/internal/cgen/base/image-impl.c index 2e688f86..3c69d3bf 100644 --- a/internal/cgen/base/image-impl.c +++ b/internal/cgen/base/image-impl.c @@ -36,8 +36,9 @@ wuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst, uint32_t b5 = 0x1F & (argb >> (8 - 5)); uint32_t g6 = 0x3F & (argb >> (16 - 6)); uint32_t r5 = 0x1F & (argb >> (24 - 5)); + uint32_t alpha = argb & 0xFF000000; wuffs_base__store_u32le__no_bounds_check( - d, (r5 << 11) | (g6 << 5) | (b5 << 0)); + d, alpha | (r5 << 11) | (g6 << 5) | (b5 << 0)); s += 4; d += 4; } @@ -450,6 +451,19 @@ wuffs_base__pixel_swizzler__copy_1_1(wuffs_base__slice_u8 dst, return wuffs_base__slice_u8__copy_from_slice(dst, src); } +static uint64_t // +wuffs_base__pixel_swizzler__copy_3_3(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t src_len3 = src.len / 3; + size_t len = dst_len3 < src_len3 ? dst_len3 : src_len3; + if (len > 0) { + memmove(dst.ptr, src.ptr, len * 3); + } + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -465,6 +479,143 @@ wuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst, // -------- +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgr(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t src_len3 = src.len / 3; + size_t len = dst_len2 < src_len3 ? dst_len2 : src_len3; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t b5 = s[0] >> 3; + uint32_t g6 = s[1] >> 2; + uint32_t r5 = s[2] >> 3; + uint32_t rgb_565 = (r5 << 11) | (g6 << 5) | (b5 << 0); + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565); + + s += 1 * 3; + d += 1 * 2; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t src_len4 = src.len / 4; + size_t len = dst_len2 < src_len4 ? dst_len2 : src_len4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + wuffs_base__store_u16le__no_bounds_check( + d + (0 * 2), + wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565( + wuffs_base__premul_u32_axxx( + wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))))); + + s += 1 * 4; + d += 1 * 2; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t src_len4 = src.len / 4; + size_t len = dst_len2 < src_len4 ? dst_len2 : src_len4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + // Convert from 8-bit color to 16-bit color. + uint32_t sa = 0x101 * ((uint32_t)s[3]); + uint32_t sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 0x101 * ((uint32_t)s[0]); + + // Convert from 565 color to 16-bit color. + uint32_t old_rgb_565 = wuffs_base__load_u16le__no_bounds_check(d + (0 * 2)); + uint32_t old_r5 = 0x1F & (old_rgb_565 >> 11); + uint32_t dr = (0x8421 * old_r5) >> 4; + uint32_t old_g6 = 0x3F & (old_rgb_565 >> 5); + uint32_t dg = (0x1041 * old_g6) >> 2; + uint32_t old_b5 = 0x1F & (old_rgb_565 >> 0); + uint32_t db = (0x8421 * old_b5) >> 4; + + // Calculate the inverse of the src-alpha: how much of the dst to keep. + uint32_t ia = 0xFFFF - sa; + + // Composite src (nonpremul) over dst (premul). + dr = ((sr * sa) + (dr * ia)) / 0xFFFF; + dg = ((sg * sa) + (dg * ia)) / 0xFFFF; + db = ((sb * sa) + (db * ia)) / 0xFFFF; + + // Convert from 16-bit color to 565 color and combine the components. + uint32_t new_r5 = 0x1F & (dr >> 11); + uint32_t new_g6 = 0x3F & (dg >> 10); + uint32_t new_b5 = 0x1F & (db >> 11); + uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5) | (new_b5 << 0); + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), + (uint16_t)new_rgb_565); + + s += 1 * 4; + d += 1 * 2; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__y(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t len = dst_len2 < src.len ? dst_len2 : src.len; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t y5 = s[0] >> 3; + uint32_t y6 = s[0] >> 2; + uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0); + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565); + + s += 1 * 1; + d += 1 * 2; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgr_565__index__src( wuffs_base__slice_u8 dst, @@ -513,6 +664,111 @@ wuffs_base__pixel_swizzler__bgr_565__index__src( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__index_binary_alpha__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + if (dst_palette.len != 1024) { + return 0; + } + size_t dst_len2 = dst.len / 2; + size_t len = dst_len2 < src.len ? dst_len2 : src.len; + 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(dst_palette.ptr + + ((size_t)s[0] * 4)); + if (s0) { + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)s0); + } + + s += 1 * 1; + d += 1 * 2; + n -= 1; + } + + return len; +} + +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t src_len4 = src.len / 4; + size_t len = dst_len3 < src_len4 ? dst_len3 : src_len4; + uint8_t* d = dst.ptr; + 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_u24le__no_bounds_check(d + (0 * 3), s0); + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t src_len4 = src.len / 4; + size_t len = dst_len3 < src_len4 ? dst_len3 : src_len4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + // Convert from 8-bit color to 16-bit color. + uint32_t sa = 0x101 * ((uint32_t)s[3]); + uint32_t sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 0x101 * ((uint32_t)s[0]); + uint32_t dr = 0x101 * ((uint32_t)d[2]); + uint32_t dg = 0x101 * ((uint32_t)d[1]); + uint32_t db = 0x101 * ((uint32_t)d[0]); + + // Calculate the inverse of the src-alpha: how much of the dst to keep. + uint32_t ia = 0xFFFF - sa; + + // Composite src (nonpremul) over dst (premul). + dr = ((sr * sa) + (dr * ia)) / 0xFFFF; + dg = ((sg * sa) + (dg * ia)) / 0xFFFF; + db = ((sb * sa) + (db * ia)) / 0xFFFF; + + // Convert from 16-bit color to 8-bit color. + d[0] = (uint8_t)(db >> 8); + d[1] = (uint8_t)(dg >> 8); + d[2] = (uint8_t)(dr >> 8); + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -715,6 +971,32 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__xxx__y(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t len = dst_len3 < src.len ? dst_len3 : src.len; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint8_t s0 = s[0]; + d[0] = s0; + d[1] = s0; + d[2] = s0; + + s += 1 * 1; + d += 1 * 3; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -882,13 +1164,11 @@ wuffs_base__pixel_swizzler__prepare__y(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_blend blend) { switch (dst_format.repr) { case WUFFS_BASE__PIXEL_FORMAT__BGR_565: - // TODO. - break; + return wuffs_base__pixel_swizzler__bgr_565__y; case WUFFS_BASE__PIXEL_FORMAT__BGR: case WUFFS_BASE__PIXEL_FORMAT__RGB: - // TODO. - break; + return wuffs_base__pixel_swizzler__xxx__y; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: @@ -932,6 +1212,8 @@ wuffs_base__pixel_swizzler__prepare__indexed__bgra_binary( switch (blend) { case WUFFS_BASE__PIXEL_BLEND__SRC: return wuffs_base__pixel_swizzler__bgr_565__index__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr_565__index_binary_alpha__src_over; } return NULL; @@ -1002,12 +1284,10 @@ wuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_blend blend) { switch (dst_format.repr) { case WUFFS_BASE__PIXEL_FORMAT__BGR_565: - // TODO. - break; + return wuffs_base__pixel_swizzler__bgr_565__bgr; case WUFFS_BASE__PIXEL_FORMAT__BGR: - // TODO. - break; + return wuffs_base__pixel_swizzler__copy_3_3; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: @@ -1035,12 +1315,22 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul( wuffs_base__pixel_blend blend) { switch (dst_format.repr) { case WUFFS_BASE__PIXEL_FORMAT__BGR_565: - // TODO. - break; + switch (blend) { + case WUFFS_BASE__PIXEL_BLEND__SRC: + return wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src_over; + } + return NULL; case WUFFS_BASE__PIXEL_FORMAT__BGR: - // TODO. - break; + switch (blend) { + case WUFFS_BASE__PIXEL_BLEND__SRC: + return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src_over; + } + return NULL; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: switch (blend) { diff --git a/internal/cgen/data.go b/internal/cgen/data.go index 3a3a10d3..fb9ca00c 100644 --- a/internal/cgen/data.go +++ b/internal/cgen/data.go @@ -29,8 +29,8 @@ 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\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 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 uint32_t alpha = argb & 0xFF000000;\n wuffs_base__store_u32le__no_bounds_check(\n d, alpha | (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 uint8_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))" + @@ -48,10 +48,19 @@ const baseImageImplC = "" + "// --------\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 (uint8_t)best_index;\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__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_3_3(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len3 = dst.len / 3;\n size_t src_len3 = src.len / 3;\n size_t len = dst_len3 < src_len3 ? dst_len3 : src_len3;\n if (len > 0) {\n memmove(dst.ptr, src.ptr, len * 3);\n }\n return len;\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__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__bgr_565__bgr(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len2 = dst.len / 2;\n size_t src_len3 = src.len / 3;\n size_t len = dst_len2 < src_len3 ? dst_len2 : src_len3;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t b5 = s[0] >> 3;\n uint32_t g6 = s[1] >> 2;\n uint32_t r5 = s[2] >> 3;\n uint32_t rgb_565 = (r5 << 11) | (g6 << 5) | (b5 << 0);\n wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565);\n\n s += 1 * 3;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__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_len2 = dst.len / 2;\n size_t src_len4 = src.len / 4;\n size_t len = " + + "dst_len2 < src_len4 ? dst_len2 : src_len4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u16le__no_bounds_check(\n d + (0 * 2),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)))));\n\n s += 1 * 4;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__bgra_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_len2 = dst.len / 2;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len2 < src_len4 ? dst_len2 : src_len4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Convert from 8-bit color to 16-bit color.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t sr = 0x101 * ((uint32_t)s[2]);\n uint32_" + + "t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sb = 0x101 * ((uint32_t)s[0]);\n\n // Convert from 565 color to 16-bit color.\n uint32_t old_rgb_565 = wuffs_base__load_u16le__no_bounds_check(d + (0 * 2));\n uint32_t old_r5 = 0x1F & (old_rgb_565 >> 11);\n uint32_t dr = (0x8421 * old_r5) >> 4;\n uint32_t old_g6 = 0x3F & (old_rgb_565 >> 5);\n uint32_t dg = (0x1041 * old_g6) >> 2;\n uint32_t old_b5 = 0x1F & (old_rgb_565 >> 0);\n uint32_t db = (0x8421 * old_b5) >> 4;\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul).\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Convert from 16-bit color to 565 color and combine the components.\n uint32_t new_r5 = 0x1F & (dr >> 11);\n uint32_t new_g6 = 0x3F & (dg >> 10);\n uint32_t new_b5 = 0x1F & (db >> 11);\n uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5" + + ") | (new_b5 << 0);\n wuffs_base__store_u16le__no_bounds_check(d + (0 * 2),\n (uint16_t)new_rgb_565);\n\n s += 1 * 4;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__y(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 // TODO: unroll.\n\n while (n >= 1) {\n uint32_t y5 = s[0] >> 3;\n uint32_t y6 = s[0] >> 2;\n uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0);\n wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565);\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__index__src(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_p" + + "alette,\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 += l" + + "oop_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__bgr_565__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_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 // TODO: unroll.\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_u16le__no_bounds_check(d + (0 * 2), (uint16_t)s0);\n " + + " }\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\n" + + "" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len3 = dst.len / 3;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len3 < src_len4 ? dst_len3 : src_len4;\n uint8_t* d = dst.ptr;\n 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_u24le__no_bounds_check(d + (0 * 3), s0);\n\n s += 1 * 4;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__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_len3 = dst.len / 3;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len3 < src_len4 ? dst_len3 : src_len4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len" + + ";\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Convert from 8-bit color to 16-bit color.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t sr = 0x101 * ((uint32_t)s[2]);\n uint32_t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sb = 0x101 * ((uint32_t)s[0]);\n uint32_t dr = 0x101 * ((uint32_t)d[2]);\n uint32_t dg = 0x101 * ((uint32_t)d[1]);\n uint32_t db = 0x101 * ((uint32_t)d[0]);\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul).\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Convert from 16-bit color to 8-bit color.\n d[0] = (uint8_t)(db >> 8);\n d[1] = (uint8_t)(dg >> 8);\n d[2] = (uint8_t)(dr >> 8);\n\n s += 1 * 4;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\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" + "" + @@ -61,7 +70,8 @@ const baseImageImplC = "" + "// --------\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 * 3), 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 * 3), 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 * 3), 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 * 3), 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 * 3), s0);\n }\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\n" + + " ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u24le__no_bounds_check(d + (2 * 3), 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 * 3), 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 * 3), 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__xxx__y(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\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 // TODO: unroll.\n\n while (n >= 1) {\n uint8_t s0 = s[0];\n d[0] = s0;\n d[1] = s0;\n d[2] = s0;\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)" + @@ -69,13 +79,14 @@ const baseImageImplC = "" + " 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__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\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 return wuffs_base__pixel_swizzler__bgr_565__y;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n return wuffs_base__pixel_swizzler__xxx__y;\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 WU" + + "FFS_BASE__PIXEL_FORMAT__RGBX:\n return wuffs_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__bgr_565__index__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr_565__index_binary_alpha__src_over;\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_pa" + + "lette) !=\n 1024) {\n return NULL;\n }\n switch (blend) {\n case WUFFS_BASE__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 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__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 return wuffs_base__pixel_swizzler__bgr_565__bgr;\n\n case WUFFS_BASE__PIXEL_FORMAT__B" + + "GR:\n return wuffs_base__pixel_swizzler__copy_3_3;\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 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 switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_b" + + "ase__pixel_swizzler__bgr_565__bgra_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n 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_swiz" + + "zler__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" + diff --git a/release/c/wuffs-unsupported-snapshot.c b/release/c/wuffs-unsupported-snapshot.c index 73d07eb4..b522e68e 100644 --- a/release/c/wuffs-unsupported-snapshot.c +++ b/release/c/wuffs-unsupported-snapshot.c @@ -8542,8 +8542,9 @@ wuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst, uint32_t b5 = 0x1F & (argb >> (8 - 5)); uint32_t g6 = 0x3F & (argb >> (16 - 6)); uint32_t r5 = 0x1F & (argb >> (24 - 5)); + uint32_t alpha = argb & 0xFF000000; wuffs_base__store_u32le__no_bounds_check( - d, (r5 << 11) | (g6 << 5) | (b5 << 0)); + d, alpha | (r5 << 11) | (g6 << 5) | (b5 << 0)); s += 4; d += 4; } @@ -8956,6 +8957,19 @@ wuffs_base__pixel_swizzler__copy_1_1(wuffs_base__slice_u8 dst, return wuffs_base__slice_u8__copy_from_slice(dst, src); } +static uint64_t // +wuffs_base__pixel_swizzler__copy_3_3(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t src_len3 = src.len / 3; + size_t len = dst_len3 < src_len3 ? dst_len3 : src_len3; + if (len > 0) { + memmove(dst.ptr, src.ptr, len * 3); + } + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst, wuffs_base__slice_u8 dst_palette, @@ -8971,6 +8985,143 @@ wuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst, // -------- +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgr(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t src_len3 = src.len / 3; + size_t len = dst_len2 < src_len3 ? dst_len2 : src_len3; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t b5 = s[0] >> 3; + uint32_t g6 = s[1] >> 2; + uint32_t r5 = s[2] >> 3; + uint32_t rgb_565 = (r5 << 11) | (g6 << 5) | (b5 << 0); + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565); + + s += 1 * 3; + d += 1 * 2; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t src_len4 = src.len / 4; + size_t len = dst_len2 < src_len4 ? dst_len2 : src_len4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + wuffs_base__store_u16le__no_bounds_check( + d + (0 * 2), + wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565( + wuffs_base__premul_u32_axxx( + wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))))); + + s += 1 * 4; + d += 1 * 2; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t src_len4 = src.len / 4; + size_t len = dst_len2 < src_len4 ? dst_len2 : src_len4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + // Convert from 8-bit color to 16-bit color. + uint32_t sa = 0x101 * ((uint32_t)s[3]); + uint32_t sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 0x101 * ((uint32_t)s[0]); + + // Convert from 565 color to 16-bit color. + uint32_t old_rgb_565 = wuffs_base__load_u16le__no_bounds_check(d + (0 * 2)); + uint32_t old_r5 = 0x1F & (old_rgb_565 >> 11); + uint32_t dr = (0x8421 * old_r5) >> 4; + uint32_t old_g6 = 0x3F & (old_rgb_565 >> 5); + uint32_t dg = (0x1041 * old_g6) >> 2; + uint32_t old_b5 = 0x1F & (old_rgb_565 >> 0); + uint32_t db = (0x8421 * old_b5) >> 4; + + // Calculate the inverse of the src-alpha: how much of the dst to keep. + uint32_t ia = 0xFFFF - sa; + + // Composite src (nonpremul) over dst (premul). + dr = ((sr * sa) + (dr * ia)) / 0xFFFF; + dg = ((sg * sa) + (dg * ia)) / 0xFFFF; + db = ((sb * sa) + (db * ia)) / 0xFFFF; + + // Convert from 16-bit color to 565 color and combine the components. + uint32_t new_r5 = 0x1F & (dr >> 11); + uint32_t new_g6 = 0x3F & (dg >> 10); + uint32_t new_b5 = 0x1F & (db >> 11); + uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5) | (new_b5 << 0); + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), + (uint16_t)new_rgb_565); + + s += 1 * 4; + d += 1 * 2; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__y(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len2 = dst.len / 2; + size_t len = dst_len2 < src.len ? dst_len2 : src.len; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t y5 = s[0] >> 3; + uint32_t y6 = s[0] >> 2; + uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0); + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565); + + s += 1 * 1; + d += 1 * 2; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgr_565__index__src( wuffs_base__slice_u8 dst, @@ -9019,6 +9170,111 @@ wuffs_base__pixel_swizzler__bgr_565__index__src( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__index_binary_alpha__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + if (dst_palette.len != 1024) { + return 0; + } + size_t dst_len2 = dst.len / 2; + size_t len = dst_len2 < src.len ? dst_len2 : src.len; + 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(dst_palette.ptr + + ((size_t)s[0] * 4)); + if (s0) { + wuffs_base__store_u16le__no_bounds_check(d + (0 * 2), (uint16_t)s0); + } + + s += 1 * 1; + d += 1 * 2; + n -= 1; + } + + return len; +} + +// -------- + +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t src_len4 = src.len / 4; + size_t len = dst_len3 < src_len4 ? dst_len3 : src_len4; + uint8_t* d = dst.ptr; + 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_u24le__no_bounds_check(d + (0 * 3), s0); + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src_over( + wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t src_len4 = src.len / 4; + size_t len = dst_len3 < src_len4 ? dst_len3 : src_len4; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + // Convert from 8-bit color to 16-bit color. + uint32_t sa = 0x101 * ((uint32_t)s[3]); + uint32_t sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 0x101 * ((uint32_t)s[0]); + uint32_t dr = 0x101 * ((uint32_t)d[2]); + uint32_t dg = 0x101 * ((uint32_t)d[1]); + uint32_t db = 0x101 * ((uint32_t)d[0]); + + // Calculate the inverse of the src-alpha: how much of the dst to keep. + uint32_t ia = 0xFFFF - sa; + + // Composite src (nonpremul) over dst (premul). + dr = ((sr * sa) + (dr * ia)) / 0xFFFF; + dg = ((sg * sa) + (dg * ia)) / 0xFFFF; + db = ((sb * sa) + (db * ia)) / 0xFFFF; + + // Convert from 16-bit color to 8-bit color. + d[0] = (uint8_t)(db >> 8); + d[1] = (uint8_t)(dg >> 8); + d[2] = (uint8_t)(dr >> 8); + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -9221,6 +9477,32 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__xxx__y(wuffs_base__slice_u8 dst, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src) { + size_t dst_len3 = dst.len / 3; + size_t len = dst_len3 < src.len ? dst_len3 : src.len; + uint8_t* d = dst.ptr; + uint8_t* s = src.ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint8_t s0 = s[0]; + d[0] = s0; + d[1] = s0; + d[2] = s0; + + s += 1 * 1; + d += 1 * 3; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -9388,13 +9670,11 @@ wuffs_base__pixel_swizzler__prepare__y(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_blend blend) { switch (dst_format.repr) { case WUFFS_BASE__PIXEL_FORMAT__BGR_565: - // TODO. - break; + return wuffs_base__pixel_swizzler__bgr_565__y; case WUFFS_BASE__PIXEL_FORMAT__BGR: case WUFFS_BASE__PIXEL_FORMAT__RGB: - // TODO. - break; + return wuffs_base__pixel_swizzler__xxx__y; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: @@ -9438,6 +9718,8 @@ wuffs_base__pixel_swizzler__prepare__indexed__bgra_binary( switch (blend) { case WUFFS_BASE__PIXEL_BLEND__SRC: return wuffs_base__pixel_swizzler__bgr_565__index__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr_565__index_binary_alpha__src_over; } return NULL; @@ -9508,12 +9790,10 @@ wuffs_base__pixel_swizzler__prepare__bgr(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_blend blend) { switch (dst_format.repr) { case WUFFS_BASE__PIXEL_FORMAT__BGR_565: - // TODO. - break; + return wuffs_base__pixel_swizzler__bgr_565__bgr; case WUFFS_BASE__PIXEL_FORMAT__BGR: - // TODO. - break; + return wuffs_base__pixel_swizzler__copy_3_3; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: @@ -9541,12 +9821,22 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul( wuffs_base__pixel_blend blend) { switch (dst_format.repr) { case WUFFS_BASE__PIXEL_FORMAT__BGR_565: - // TODO. - break; + switch (blend) { + case WUFFS_BASE__PIXEL_BLEND__SRC: + return wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src_over; + } + return NULL; case WUFFS_BASE__PIXEL_FORMAT__BGR: - // TODO. - break; + switch (blend) { + case WUFFS_BASE__PIXEL_BLEND__SRC: + return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src_over; + } + return NULL; case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: switch (blend) { diff --git a/test/c/std/wbmp.c b/test/c/std/wbmp.c index 34705244..0cb7ad49 100644 --- a/test/c/std/wbmp.c +++ b/test/c/std/wbmp.c @@ -136,10 +136,22 @@ test_wuffs_pixel_swizzler_swizzle() { wuffs_base__color_u32_argb_premul pixel; uint32_t pixfmt_repr; } dsts[] = { + { + .pixel = 0xFF000010, + .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGR_565, + }, + { + .pixel = 0xFF000040, + .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGR, + }, { .pixel = 0x80000040, .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL, }, + { + .pixel = 0x80000040, + .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL, + }, }; const wuffs_base__pixel_blend blends[] = { @@ -184,6 +196,10 @@ test_wuffs_pixel_swizzler_swizzle() { wuffs_base__pixel_buffer__set_from_slice( &dst_pixbuf, &dst_pixcfg, g_have_slice_u8)); fill_palette_with_grays(&dst_pixbuf); + wuffs_base__pixel_format dst_pixfmt = + wuffs_base__make_pixel_format(dsts[d].pixfmt_repr); + wuffs_base__pixel_alpha_transparency dst_transparency = + wuffs_base__pixel_format__transparency(&dst_pixfmt); wuffs_base__slice_u8 dst_palette = wuffs_base__pixel_buffer__palette(&dst_pixbuf); @@ -204,8 +220,8 @@ test_wuffs_pixel_swizzler_swizzle() { CHECK_STATUS( "prepare", wuffs_base__pixel_swizzler__prepare( - &swizzler, wuffs_base__make_pixel_format(dsts[d].pixfmt_repr), - dst_palette, wuffs_base__make_pixel_format(srcs[s].pixfmt_repr), + &swizzler, dst_pixfmt, dst_palette, + wuffs_base__make_pixel_format(srcs[s].pixfmt_repr), wuffs_base__pixel_buffer__palette(&src_pixbuf), blends[b])); wuffs_base__pixel_swizzler__swizzle_interleaved( &swizzler, @@ -216,17 +232,21 @@ test_wuffs_pixel_swizzler_swizzle() { wuffs_base__pixel_buffer__plane(&src_pixbuf, 0), height / 2)); // Check the middle dst pixel. - uint32_t tolerance = 0; + uint32_t tolerance = + (dsts[d].pixfmt_repr == WUFFS_BASE__PIXEL_FORMAT__BGR_565) ? 4 : 0; wuffs_base__color_u32_argb_premul want_dst_pixel = 0; if (blends[b] == WUFFS_BASE__PIXEL_BLEND__SRC) { want_dst_pixel = srcs[s].pixel; } else if (blends[b] == WUFFS_BASE__PIXEL_BLEND__SRC_OVER) { - tolerance = 1; + tolerance += 1; want_dst_pixel = wuffs_base__composite_premul_premul_u32_axxx( dsts[d].pixel, srcs[s].pixel); } else { return "unsupported blend"; } + if (dst_transparency == WUFFS_BASE__PIXEL_ALPHA_TRANSPARENCY__OPAQUE) { + want_dst_pixel |= 0xFF000000; + } wuffs_base__color_u32_argb_premul have_dst_pixel = wuffs_base__pixel_buffer__color_u32_at(&dst_pixbuf, width / 2, height / 2);