diff --git a/internal/cgen/base/pixconv-submodule.c b/internal/cgen/base/pixconv-submodule.c index 2cf92de9..4298b1f1 100644 --- a/internal/cgen/base/pixconv-submodule.c +++ b/internal/cgen/base/pixconv-submodule.c @@ -634,6 +634,37 @@ wuffs_base__pixel_swizzler__bgr_565__bgr(uint8_t* dst_ptr, return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgrx(uint8_t* dst_ptr, + size_t dst_len, + uint8_t* dst_palette_ptr, + size_t dst_palette_len, + const uint8_t* src_ptr, + size_t src_len) { + size_t dst_len2 = dst_len / 2; + size_t src_len4 = src_len / 4; + size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + 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 * 4; + d += 1 * 2; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src( uint8_t* dst_ptr, @@ -1279,6 +1310,37 @@ wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src_over( // -------- +static uint64_t // +wuffs_base__pixel_swizzler__bgrw__bgrx(uint8_t* dst_ptr, + size_t dst_len, + uint8_t* dst_palette_ptr, + size_t dst_palette_len, + const uint8_t* src_ptr, + size_t src_len) { + size_t dst_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + wuffs_base__store_u32le__no_bounds_check( + d + (0 * 4), + 0xFF000000 | wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +// -------- + static uint64_t // wuffs_base__pixel_swizzler__xxx__index__src(uint8_t* dst_ptr, size_t dst_len, @@ -1397,6 +1459,34 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__xxx__xxxx(uint8_t* dst_ptr, + size_t dst_len, + uint8_t* dst_palette_ptr, + size_t dst_palette_len, + const uint8_t* src_ptr, + size_t src_len) { + size_t dst_len3 = dst_len / 3; + size_t src_len4 = src_len / 4; + size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + wuffs_base__store_u24le__no_bounds_check( + d + (0 * 3), wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))); + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__xxx__y(uint8_t* dst_ptr, size_t dst_len, @@ -1867,6 +1957,38 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le( return NULL; } +static wuffs_base__pixel_swizzler__func // +wuffs_base__pixel_swizzler__prepare__bgrx(wuffs_base__pixel_swizzler* p, + wuffs_base__pixel_format dst_pixfmt, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src_palette, + wuffs_base__pixel_blend blend) { + switch (dst_pixfmt.repr) { + case WUFFS_BASE__PIXEL_FORMAT__BGR_565: + return wuffs_base__pixel_swizzler__bgr_565__bgrx; + + case WUFFS_BASE__PIXEL_FORMAT__BGR: + return wuffs_base__pixel_swizzler__xxx__xxxx; + + case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: + case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: + case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY: + return wuffs_base__pixel_swizzler__bgrw__bgrx; + + case WUFFS_BASE__PIXEL_FORMAT__BGRX: + return wuffs_base__pixel_swizzler__copy_4_4; + + 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: + case WUFFS_BASE__PIXEL_FORMAT__RGBX: + // TODO. + break; + } + return NULL; +} + // -------- WUFFS_BASE__MAYBE_STATIC wuffs_base__status // @@ -1918,6 +2040,11 @@ wuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p, func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le( p, dst_pixfmt, dst_palette, src_palette, blend); break; + + case WUFFS_BASE__PIXEL_FORMAT__BGRX: + func = wuffs_base__pixel_swizzler__prepare__bgrx( + p, dst_pixfmt, dst_palette, src_palette, blend); + break; } p->private_impl.func = func; diff --git a/internal/cgen/data/data.go b/internal/cgen/data/data.go index 41b54ba5..ad142669 100644 --- a/internal/cgen/data/data.go +++ b/internal/cgen/data/data.go @@ -545,16 +545,17 @@ const BasePixConvSubmoduleC = "" + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_1_1(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t len = (dst_len < src_len) ? dst_len : src_len;\n if (len > 0) {\n memmove(dst_ptr, src_ptr, len);\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_3_3(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_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(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4;\n 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__bgr(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len3 = src_len / 3;\n size_t len = (dst_len2 < src_len3) ? dst_len2 : src_len3;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t 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 uint8_t*" + - " dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u16le__no_bounds_check(\n d + (0 * 2),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(\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_4x16le__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len8 = src_len / " + - "8;\n size_t len = (dst_len2 < src_len8) ? dst_len2 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n 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__color_u64_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__load_u64le__no_bounds_check(s + (0 * 8)))));\n\n s += 1 * 8;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Extract 16-bit color " + - "components.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t 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__bgra_nonpremul_4x16le__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len2 < src_len8) ? dst_len2 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Extract 16-bit color components.\n uint32_t sa = ((uint32_t)wuffs_base__load_u16le__no_bounds_check(s + 6));\n uint32_t sr = ((uint32_t)wuffs_base__load_u16le__no_bounds_check(s + 4));\n uint32_t sg = ((uint32" + - "_t)wuffs_base__load_u16le__no_bounds_check(s + 2));\n uint32_t sb = ((uint32_t)wuffs_base__load_u16le__no_bounds_check(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 * 8;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__y(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t y5 = s[0] >> 3;\n uint32_t y6 = s[0] >> 2;\n uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0);\n wuffs_base__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(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__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_ch" + - "eck(\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__bgr_565__index_binary_alpha__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n " + - " const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__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_565__bgr(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len3 = src_len / 3;\n size_t len = (dst_len2 < src_len3) ? dst_len2 : src_len3;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t 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__bgrx(uint8_t* dst_ptr,\n " + + " size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n 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 * 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(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const" + + " uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u16le__no_bounds_check(\n d + (0 * 2),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(\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_4x16le__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len2 < src_len8) ? dst_len2 : src_len8;\n uint8_t* d = dst_ptr;\n const ui" + + "nt8_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__color_u64_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__load_u64le__no_bounds_check(s + (0 * 8)))));\n\n s += 1 * 8;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t 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__bgra_nonpremul_4x16le__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len2 < src_len8) ? dst_len2 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Extract 16-bit color components.\n uint32_t sa = ((uint32_t)wuffs_base__load_u16le__no_bounds_check(s + 6));\n uint32_t sr = ((uint32_t)wuffs_base__load_u16le__no_bounds_check(s + 4));\n uint32_t sg = ((uint32_t)wuffs_base__load_u16le__no_bounds_check(s + 2));\n uint32_t sb = ((uint32_t)wuffs_base__load" + + "_u16le__no_bounds_check(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_u16" + + "le__no_bounds_check(d + (0 * 2),\n (uint16_t)new_rgb_565);\n\n s += 1 * 8;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__y(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t y5 = s[0] >> 3;\n uint32_t y6 = s[0] >> 2;\n uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0);\n wuffs_base__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 uin" + + "t64_t //\nwuffs_base__pixel_swizzler__bgr_565__index__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__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_p" + + "alette_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__bgr_565__index_binary_alpha__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n " + + "}\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__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(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 =\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__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 //\nw" + "uffs_base__pixel_swizzler__bgr__bgra_nonpremul_4x16le__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len3 < src_len8) ? dst_len3 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 =\n wuffs_base__color_u64_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__load_u64le__no_bounds_check(s + (0 * 8)));\n wuffs_base__store_u24le__no_bounds_check(d + (0 * 3), s0);\n\n s += 1 * 8;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_le" + @@ -570,11 +571,14 @@ const BasePixConvSubmoduleC = "" + "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__bgra_premul__bgra_nonpremul_4x16le__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len4 < src_len8) ? dst_len4 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint64_t d0 = wuffs_base__color_u32__as__color_u64(\n wuffs_base__load_u32le__no_bounds_check(d + (0 * 4)));\n uint64_t s0 = wuffs_base__load_u64le__no_bounds_check(s + (0 * 8));\n wuffs_base__store_u32le__no_bounds_check(\n d + (0" + " * 4),\n wuffs_base__color_u64__as__color_u32(\n wuffs_base__composite_premul_nonpremul_u64_axxx(d0, s0)));\n\n s += 1 * 8;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" + "" + + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgrw__bgrx(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4),\n 0xFF000000 | wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" + + "" + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__index__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst_len / 3;\n size_t len = (dst_len3 < src_len) ? dst_len3 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n // 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_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 ((siz" + "e_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 uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst_len / 3;\n size_t len = (dst_len3 < src_len) ? dst_len3 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n 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\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__y(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t len = (dst_len3 < src_len) ? dst_len3 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n 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}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__xxxx(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u24le__no_bounds_check(\n d + (0 * 3), wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)));\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__xxx__y(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n " + + " size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t len = (dst_len3 < src_len) ? dst_len3 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n 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(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len4 = dst_len / 4;\n size_t len = (dst_len4 < src_len) ? dst_len4 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n 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 uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n" + @@ -592,12 +596,13 @@ const BasePixConvSubmoduleC = "" + "premul__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_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\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le(\n wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_pixfmt,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel" + "_swizzler__bgr_565__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul_4x16le__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul_4x16le__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__squash_tight_4x8_4x16le;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul_4x16le__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:" + - "\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src_over;\n }\n return NULL;\n\n case 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 return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src_over;\n }\n return NULL;\n\n case 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\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgrx(wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_pixfmt,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n " + + " wuffs_base__pixel_blend blend) {\n switch (dst_pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n return wuffs_base__pixel_swizzler__bgr_565__bgrx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return wuffs_base__pixel_swizzler__xxx__xxxx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n return wuffs_base__pixel_swizzler__bgrw__bgrx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n return wuffs_base__pixel_swizzler__copy_4_4;\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__MAYBE_STATIC wuffs_base__status //\nwuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_pixfmt,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__pixel_format src_pixfmt,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n if (!p) {\n return wuffs_base__make_status(wuffs_base__error__bad_receiver);\n }\n p->private_impl.func = NULL;\n p->private_impl.src_pixfmt_bytes_per_pixel = 0;\n\n wuffs_base__pixel_swizzler__func func = NULL;\n uint32_t src_pixfmt_bits_per_pixel =\n wuffs_base__pixel_format__bits_per_pixel(&src_pixfmt);\n if ((src_pixfmt_bits_per_pixel == 0) ||\n ((src_pixfmt_bits_per_pixel & 7) != 0)) {\n return wuffs_base__make_status(\n wuffs_base__error__unsupported_pixel_swizzler_option);\n }\n\n // TODO: support many more for" + - "mats.\n\n switch (src_pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n func = wuffs_base__pixel_swizzler__prepare__y(p, dst_pixfmt, dst_palette,\n src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY:\n func = wuffs_base__pixel_swizzler__prepare__indexed__bgra_binary(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n func = wuffs_base__pixel_swizzler__prepare__bgr(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n }\n\n p->private_im" + - "pl.func = func;\n p->private_impl.src_pixfmt_bytes_per_pixel = src_pixfmt_bits_per_pixel / 8;\n return wuffs_base__make_status(\n func ? NULL : wuffs_base__error__unsupported_pixel_swizzler_option);\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_reader(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n const uint8_t** ptr_iop_r,\n const uint8_t* io2_r) {\n if (p && p->private_impl.func) {\n const uint8_t* iop_r = *ptr_iop_r;\n uint64_t n = (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, iop_r,\n (size_t)(io2_r - iop_r));\n *ptr_iop_r += n * p->private_impl.src_pixfmt_bytes_per_pixel;\n return n;\n }\n return 0;\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_slice(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n " + - " wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (p && p->private_impl.func) {\n return (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, src.ptr, src.len);\n }\n return 0;\n}\n" + + "mats.\n\n switch (src_pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n func = wuffs_base__pixel_swizzler__prepare__y(p, dst_pixfmt, dst_palette,\n src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY:\n func = wuffs_base__pixel_swizzler__prepare__indexed__bgra_binary(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n func = wuffs_base__pixel_swizzler__prepare__bgr(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE" + + "__PIXEL_FORMAT__BGRX:\n func = wuffs_base__pixel_swizzler__prepare__bgrx(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n }\n\n p->private_impl.func = func;\n p->private_impl.src_pixfmt_bytes_per_pixel = src_pixfmt_bits_per_pixel / 8;\n return wuffs_base__make_status(\n func ? NULL : wuffs_base__error__unsupported_pixel_swizzler_option);\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_reader(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n const uint8_t** ptr_iop_r,\n const uint8_t* io2_r) {\n if (p && p->private_impl.func) {\n const uint8_t* iop_r = *ptr_iop_r;\n uint64_t n = (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, iop_r,\n (size_t)(io2_r - iop_r));\n *ptr_iop_r += n * p->private_impl.src_pixfmt_bytes_per_pixel;\n return n;\n }\n return 0;\n}\n\n" + + "WUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_slice(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (p && p->private_impl.func) {\n return (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, src.ptr, src.len);\n }\n return 0;\n}\n" + "" const BaseUTF8SubmoduleC = "" + diff --git a/release/c/wuffs-unsupported-snapshot.c b/release/c/wuffs-unsupported-snapshot.c index 1fc900eb..9ace6d50 100644 --- a/release/c/wuffs-unsupported-snapshot.c +++ b/release/c/wuffs-unsupported-snapshot.c @@ -14090,6 +14090,37 @@ wuffs_base__pixel_swizzler__bgr_565__bgr(uint8_t* dst_ptr, return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgrx(uint8_t* dst_ptr, + size_t dst_len, + uint8_t* dst_palette_ptr, + size_t dst_palette_len, + const uint8_t* src_ptr, + size_t src_len) { + size_t dst_len2 = dst_len / 2; + size_t src_len4 = src_len / 4; + size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + 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 * 4; + d += 1 * 2; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul__src( uint8_t* dst_ptr, @@ -14735,6 +14766,37 @@ wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src_over( // -------- +static uint64_t // +wuffs_base__pixel_swizzler__bgrw__bgrx(uint8_t* dst_ptr, + size_t dst_len, + uint8_t* dst_palette_ptr, + size_t dst_palette_len, + const uint8_t* src_ptr, + size_t src_len) { + size_t dst_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + wuffs_base__store_u32le__no_bounds_check( + d + (0 * 4), + 0xFF000000 | wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +// -------- + static uint64_t // wuffs_base__pixel_swizzler__xxx__index__src(uint8_t* dst_ptr, size_t dst_len, @@ -14853,6 +14915,34 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__xxx__xxxx(uint8_t* dst_ptr, + size_t dst_len, + uint8_t* dst_palette_ptr, + size_t dst_palette_len, + const uint8_t* src_ptr, + size_t src_len) { + size_t dst_len3 = dst_len / 3; + size_t src_len4 = src_len / 4; + size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + wuffs_base__store_u24le__no_bounds_check( + d + (0 * 3), wuffs_base__load_u32le__no_bounds_check(s + (0 * 4))); + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__xxx__y(uint8_t* dst_ptr, size_t dst_len, @@ -15323,6 +15413,38 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le( return NULL; } +static wuffs_base__pixel_swizzler__func // +wuffs_base__pixel_swizzler__prepare__bgrx(wuffs_base__pixel_swizzler* p, + wuffs_base__pixel_format dst_pixfmt, + wuffs_base__slice_u8 dst_palette, + wuffs_base__slice_u8 src_palette, + wuffs_base__pixel_blend blend) { + switch (dst_pixfmt.repr) { + case WUFFS_BASE__PIXEL_FORMAT__BGR_565: + return wuffs_base__pixel_swizzler__bgr_565__bgrx; + + case WUFFS_BASE__PIXEL_FORMAT__BGR: + return wuffs_base__pixel_swizzler__xxx__xxxx; + + case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL: + case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: + case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY: + return wuffs_base__pixel_swizzler__bgrw__bgrx; + + case WUFFS_BASE__PIXEL_FORMAT__BGRX: + return wuffs_base__pixel_swizzler__copy_4_4; + + 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: + case WUFFS_BASE__PIXEL_FORMAT__RGBX: + // TODO. + break; + } + return NULL; +} + // -------- WUFFS_BASE__MAYBE_STATIC wuffs_base__status // @@ -15374,6 +15496,11 @@ wuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p, func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le( p, dst_pixfmt, dst_palette, src_palette, blend); break; + + case WUFFS_BASE__PIXEL_FORMAT__BGRX: + func = wuffs_base__pixel_swizzler__prepare__bgrx( + p, dst_pixfmt, dst_palette, src_palette, blend); + break; } p->private_impl.func = func; diff --git a/test/c/std/wbmp.c b/test/c/std/wbmp.c index 3308adde..bef5ff01 100644 --- a/test/c/std/wbmp.c +++ b/test/c/std/wbmp.c @@ -134,6 +134,10 @@ test_wuffs_pixel_swizzler_swizzle() { .pixel = 0x55443300, .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE, }, + { + .pixel = 0xFF443300, + .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGRX, + }, }; const struct {