Make wuffs_base__{load,store}_uNNxe public

This commit is contained in:
Nigel Tao committed 2020-02-27 10:58:14 +11:00
1 parent 8cd45054fa
commit 1ad5cfac4d
11 files changed
+926 -856

No files matched your search

-244
View File
@@ -73,252 +73,8 @@ wuffs_base__ignore_status(wuffs_base__status z) {
#define WUFFS_BASE__UNLIKELY(expr) (expr)
#endif
// The helpers below are functions, instead of macros, because their arguments
// can be an expression that we shouldn't evaluate more than once.
//
// They are static, so that linking multiple wuffs .o files won't complain about
// duplicate function definitions.
//
// They are explicitly marked inline, even if modern compilers don't use the
// inline attribute to guide optimizations such as inlining, to avoid the
// -Wunused-function warning, and we like to compile with -Wall -Werror.
// ---------------- Numeric Types
static inline uint8_t //
wuffs_base__load_u8be(uint8_t* p) {
return p[0];
}
static inline uint16_t //
wuffs_base__load_u16be(uint8_t* p) {
return (uint16_t)(((uint16_t)(p[0]) << 8) | ((uint16_t)(p[1]) << 0));
}
static inline uint16_t //
wuffs_base__load_u16le(uint8_t* p) {
return (uint16_t)(((uint16_t)(p[0]) << 0) | ((uint16_t)(p[1]) << 8));
}
static inline uint32_t //
wuffs_base__load_u24be(uint8_t* p) {
return ((uint32_t)(p[0]) << 16) | ((uint32_t)(p[1]) << 8) |
((uint32_t)(p[2]) << 0);
}
static inline uint32_t //
wuffs_base__load_u24le(uint8_t* p) {
return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]) << 8) |
((uint32_t)(p[2]) << 16);
}
static inline uint32_t //
wuffs_base__load_u32be(uint8_t* p) {
return ((uint32_t)(p[0]) << 24) | ((uint32_t)(p[1]) << 16) |
((uint32_t)(p[2]) << 8) | ((uint32_t)(p[3]) << 0);
}
static inline uint32_t //
wuffs_base__load_u32le(uint8_t* p) {
return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]) << 8) |
((uint32_t)(p[2]) << 16) | ((uint32_t)(p[3]) << 24);
}
static inline uint64_t //
wuffs_base__load_u40be(uint8_t* p) {
return ((uint64_t)(p[0]) << 32) | ((uint64_t)(p[1]) << 24) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 8) |
((uint64_t)(p[4]) << 0);
}
static inline uint64_t //
wuffs_base__load_u40le(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32);
}
static inline uint64_t //
wuffs_base__load_u48be(uint8_t* p) {
return ((uint64_t)(p[0]) << 40) | ((uint64_t)(p[1]) << 32) |
((uint64_t)(p[2]) << 24) | ((uint64_t)(p[3]) << 16) |
((uint64_t)(p[4]) << 8) | ((uint64_t)(p[5]) << 0);
}
static inline uint64_t //
wuffs_base__load_u48le(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40);
}
static inline uint64_t //
wuffs_base__load_u56be(uint8_t* p) {
return ((uint64_t)(p[0]) << 48) | ((uint64_t)(p[1]) << 40) |
((uint64_t)(p[2]) << 32) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 16) | ((uint64_t)(p[5]) << 8) |
((uint64_t)(p[6]) << 0);
}
static inline uint64_t //
wuffs_base__load_u56le(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |
((uint64_t)(p[6]) << 48);
}
static inline uint64_t //
wuffs_base__load_u64be(uint8_t* p) {
return ((uint64_t)(p[0]) << 56) | ((uint64_t)(p[1]) << 48) |
((uint64_t)(p[2]) << 40) | ((uint64_t)(p[3]) << 32) |
((uint64_t)(p[4]) << 24) | ((uint64_t)(p[5]) << 16) |
((uint64_t)(p[6]) << 8) | ((uint64_t)(p[7]) << 0);
}
static inline uint64_t //
wuffs_base__load_u64le(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |
((uint64_t)(p[6]) << 48) | ((uint64_t)(p[7]) << 56);
}
// --------
static inline void //
wuffs_base__store_u8be(uint8_t* p, uint8_t x) {
p[0] = x;
}
static inline void //
wuffs_base__store_u16be(uint8_t* p, uint16_t x) {
p[0] = (uint8_t)(x >> 8);
p[1] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u16le(uint8_t* p, uint16_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
}
static inline void //
wuffs_base__store_u24be(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 16);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u24le(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
}
static inline void //
wuffs_base__store_u32be(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 24);
p[1] = (uint8_t)(x >> 16);
p[2] = (uint8_t)(x >> 8);
p[3] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u32le(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
}
static inline void //
wuffs_base__store_u40be(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 32);
p[1] = (uint8_t)(x >> 24);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 8);
p[4] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u40le(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
}
static inline void //
wuffs_base__store_u48be(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 40);
p[1] = (uint8_t)(x >> 32);
p[2] = (uint8_t)(x >> 24);
p[3] = (uint8_t)(x >> 16);
p[4] = (uint8_t)(x >> 8);
p[5] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u48le(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
p[5] = (uint8_t)(x >> 40);
}
static inline void //
wuffs_base__store_u56be(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 48);
p[1] = (uint8_t)(x >> 40);
p[2] = (uint8_t)(x >> 32);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 16);
p[5] = (uint8_t)(x >> 8);
p[6] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u56le(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
p[5] = (uint8_t)(x >> 40);
p[6] = (uint8_t)(x >> 48);
}
static inline void //
wuffs_base__store_u64be(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 56);
p[1] = (uint8_t)(x >> 48);
p[2] = (uint8_t)(x >> 40);
p[3] = (uint8_t)(x >> 32);
p[4] = (uint8_t)(x >> 24);
p[5] = (uint8_t)(x >> 16);
p[6] = (uint8_t)(x >> 8);
p[7] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u64le(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
p[5] = (uint8_t)(x >> 40);
p[6] = (uint8_t)(x >> 48);
p[7] = (uint8_t)(x >> 56);
}
// --------
extern const uint8_t wuffs_base__low_bits_mask__u8[9];
extern const uint16_t wuffs_base__low_bits_mask__u16[17];
extern const uint32_t wuffs_base__low_bits_mask__u32[33];
+254
View File
@@ -268,6 +268,16 @@ typedef int64_t wuffs_base__flicks;
// ---------------- Numeric Types
// The helpers below are functions, instead of macros, because their arguments
// can be an expression that we shouldn't evaluate more than once.
//
// They are static, so that linking multiple wuffs .o files won't complain about
// duplicate function definitions.
//
// They are explicitly marked inline, even if modern compilers don't use the
// inline attribute to guide optimizations such as inlining, to avoid the
// -Wunused-function warning, and we like to compile with -Wall -Werror.
static inline uint8_t //
wuffs_base__u8__min(uint8_t x, uint8_t y) {
return x < y ? x : y;
@@ -372,6 +382,250 @@ wuffs_base__u64__sat_sub(uint64_t x, uint64_t y) {
return res;
}
// --------
#define wuffs_base__load_u8be__no_bounds_check \
wuffs_base__load_u8__no_bounds_check
#define wuffs_base__load_u8le__no_bounds_check \
wuffs_base__load_u8__no_bounds_check
static inline uint8_t //
wuffs_base__load_u8__no_bounds_check(uint8_t* p) {
return p[0];
}
static inline uint16_t //
wuffs_base__load_u16be__no_bounds_check(uint8_t* p) {
return (uint16_t)(((uint16_t)(p[0]) << 8) | ((uint16_t)(p[1]) << 0));
}
static inline uint16_t //
wuffs_base__load_u16le__no_bounds_check(uint8_t* p) {
return (uint16_t)(((uint16_t)(p[0]) << 0) | ((uint16_t)(p[1]) << 8));
}
static inline uint32_t //
wuffs_base__load_u24be__no_bounds_check(uint8_t* p) {
return ((uint32_t)(p[0]) << 16) | ((uint32_t)(p[1]) << 8) |
((uint32_t)(p[2]) << 0);
}
static inline uint32_t //
wuffs_base__load_u24le__no_bounds_check(uint8_t* p) {
return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]) << 8) |
((uint32_t)(p[2]) << 16);
}
static inline uint32_t //
wuffs_base__load_u32be__no_bounds_check(uint8_t* p) {
return ((uint32_t)(p[0]) << 24) | ((uint32_t)(p[1]) << 16) |
((uint32_t)(p[2]) << 8) | ((uint32_t)(p[3]) << 0);
}
static inline uint32_t //
wuffs_base__load_u32le__no_bounds_check(uint8_t* p) {
return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]) << 8) |
((uint32_t)(p[2]) << 16) | ((uint32_t)(p[3]) << 24);
}
static inline uint64_t //
wuffs_base__load_u40be__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 32) | ((uint64_t)(p[1]) << 24) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 8) |
((uint64_t)(p[4]) << 0);
}
static inline uint64_t //
wuffs_base__load_u40le__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32);
}
static inline uint64_t //
wuffs_base__load_u48be__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 40) | ((uint64_t)(p[1]) << 32) |
((uint64_t)(p[2]) << 24) | ((uint64_t)(p[3]) << 16) |
((uint64_t)(p[4]) << 8) | ((uint64_t)(p[5]) << 0);
}
static inline uint64_t //
wuffs_base__load_u48le__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40);
}
static inline uint64_t //
wuffs_base__load_u56be__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 48) | ((uint64_t)(p[1]) << 40) |
((uint64_t)(p[2]) << 32) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 16) | ((uint64_t)(p[5]) << 8) |
((uint64_t)(p[6]) << 0);
}
static inline uint64_t //
wuffs_base__load_u56le__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |
((uint64_t)(p[6]) << 48);
}
static inline uint64_t //
wuffs_base__load_u64be__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 56) | ((uint64_t)(p[1]) << 48) |
((uint64_t)(p[2]) << 40) | ((uint64_t)(p[3]) << 32) |
((uint64_t)(p[4]) << 24) | ((uint64_t)(p[5]) << 16) |
((uint64_t)(p[6]) << 8) | ((uint64_t)(p[7]) << 0);
}
static inline uint64_t //
wuffs_base__load_u64le__no_bounds_check(uint8_t* p) {
return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |
((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |
((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |
((uint64_t)(p[6]) << 48) | ((uint64_t)(p[7]) << 56);
}
// --------
#define wuffs_base__store_u8be__no_bounds_check \
wuffs_base__store_u8__no_bounds_check
#define wuffs_base__store_u8le__no_bounds_check \
wuffs_base__store_u8__no_bounds_check
static inline void //
wuffs_base__store_u8__no_bounds_check(uint8_t* p, uint8_t x) {
p[0] = x;
}
static inline void //
wuffs_base__store_u16be__no_bounds_check(uint8_t* p, uint16_t x) {
p[0] = (uint8_t)(x >> 8);
p[1] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u16le__no_bounds_check(uint8_t* p, uint16_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
}
static inline void //
wuffs_base__store_u24be__no_bounds_check(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 16);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u24le__no_bounds_check(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
}
static inline void //
wuffs_base__store_u32be__no_bounds_check(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 24);
p[1] = (uint8_t)(x >> 16);
p[2] = (uint8_t)(x >> 8);
p[3] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u32le__no_bounds_check(uint8_t* p, uint32_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
}
static inline void //
wuffs_base__store_u40be__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 32);
p[1] = (uint8_t)(x >> 24);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 8);
p[4] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u40le__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
}
static inline void //
wuffs_base__store_u48be__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 40);
p[1] = (uint8_t)(x >> 32);
p[2] = (uint8_t)(x >> 24);
p[3] = (uint8_t)(x >> 16);
p[4] = (uint8_t)(x >> 8);
p[5] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u48le__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
p[5] = (uint8_t)(x >> 40);
}
static inline void //
wuffs_base__store_u56be__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 48);
p[1] = (uint8_t)(x >> 40);
p[2] = (uint8_t)(x >> 32);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 16);
p[5] = (uint8_t)(x >> 8);
p[6] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u56le__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
p[5] = (uint8_t)(x >> 40);
p[6] = (uint8_t)(x >> 48);
}
static inline void //
wuffs_base__store_u64be__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 56);
p[1] = (uint8_t)(x >> 48);
p[2] = (uint8_t)(x >> 40);
p[3] = (uint8_t)(x >> 32);
p[4] = (uint8_t)(x >> 24);
p[5] = (uint8_t)(x >> 16);
p[6] = (uint8_t)(x >> 8);
p[7] = (uint8_t)(x >> 0);
}
static inline void //
wuffs_base__store_u64le__no_bounds_check(uint8_t* p, uint64_t x) {
p[0] = (uint8_t)(x >> 0);
p[1] = (uint8_t)(x >> 8);
p[2] = (uint8_t)(x >> 16);
p[3] = (uint8_t)(x >> 24);
p[4] = (uint8_t)(x >> 32);
p[5] = (uint8_t)(x >> 40);
p[6] = (uint8_t)(x >> 48);
p[7] = (uint8_t)(x >> 56);
}
// ---------------- Slices and Tables
// WUFFS_BASE__SLICE is a 1-dimensional buffer.
+119 -98
View File
@@ -124,13 +124,14 @@ wuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* b,
case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:
WUFFS_BASE__FALLTHROUGH;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:
return wuffs_base__load_u32le(row + (4 * ((size_t)x)));
return wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));
case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_PREMUL:
WUFFS_BASE__FALLTHROUGH;
case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY: {
uint8_t* palette = b->private_impl.planes[3].ptr;
return wuffs_base__load_u32le(palette + (4 * ((size_t)row[x])));
return wuffs_base__load_u32le__no_bounds_check(palette +
(4 * ((size_t)row[x])));
}
// Common formats above. Rarer formats below.
@@ -141,11 +142,13 @@ wuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* b,
case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL: {
uint8_t* palette = b->private_impl.planes[3].ptr;
return wuffs_base__premul_u32_axxx(
wuffs_base__load_u32le(palette + (4 * ((size_t)row[x]))));
wuffs_base__load_u32le__no_bounds_check(palette +
(4 * ((size_t)row[x]))));
}
case WUFFS_BASE__PIXEL_FORMAT__BGR_565: {
uint16_t bgr = wuffs_base__load_u16le(row + (2 * ((size_t)x)));
uint16_t bgr =
wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x)));
uint32_t b5 = 0x1F & (bgr >> 0);
uint32_t b = (b5 << 3) | (b5 >> 2);
uint32_t g6 = 0x3F & (bgr >> 5);
@@ -155,27 +158,31 @@ wuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_buffer* b,
return 0xFF000000 | (r << 16) | (g << 8) | (b << 0);
}
case WUFFS_BASE__PIXEL_FORMAT__BGR:
return 0xFF000000 | wuffs_base__load_u24le(row + (3 * ((size_t)x)));
return 0xFF000000 |
wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x)));
case WUFFS_BASE__PIXEL_FORMAT__BGRX:
return 0xFF000000 | wuffs_base__load_u32le(row + (4 * ((size_t)x)));
return 0xFF000000 |
wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:
return wuffs_base__premul_u32_axxx(
wuffs_base__load_u32le(row + (4 * ((size_t)x))));
wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));
case WUFFS_BASE__PIXEL_FORMAT__RGB:
return wuffs_base__swap_u32_argb_abgr(
0xFF000000 | wuffs_base__load_u24le(row + (3 * ((size_t)x))));
0xFF000000 |
wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x))));
case WUFFS_BASE__PIXEL_FORMAT__RGBX:
return wuffs_base__swap_u32_argb_abgr(
0xFF000000 | wuffs_base__load_u32le(row + (4 * ((size_t)x))));
0xFF000000 |
wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));
case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:
return wuffs_base__swap_u32_argb_abgr(wuffs_base__premul_u32_axxx(
wuffs_base__load_u32le(row + (4 * ((size_t)x)))));
wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)))));
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
WUFFS_BASE__FALLTHROUGH;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:
return wuffs_base__swap_u32_argb_abgr(
wuffs_base__load_u32le(row + (4 * ((size_t)x))));
wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));
default:
// TODO: support more formats.
@@ -211,7 +218,7 @@ wuffs_base__pixel_buffer__set_color_u32_at(
case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:
WUFFS_BASE__FALLTHROUGH;
case WUFFS_BASE__PIXEL_FORMAT__BGRX:
wuffs_base__store_u32le(row + (4 * ((size_t)x)), color);
wuffs_base__store_u32le__no_bounds_check(row + (4 * ((size_t)x)), color);
break;
// Common formats above. Rarer formats below.
@@ -221,31 +228,32 @@ wuffs_base__pixel_buffer__set_color_u32_at(
uint32_t g6 = 0x3F & (color >> (16 - 6));
uint32_t r5 = 0x1F & (color >> (24 - 5));
uint32_t bgr565 = (b5 << 0) | (g6 << 5) | (r5 << 11);
wuffs_base__store_u16le(row + (2 * ((size_t)x)), (uint16_t)bgr565);
wuffs_base__store_u16le__no_bounds_check(row + (2 * ((size_t)x)),
(uint16_t)bgr565);
break;
}
case WUFFS_BASE__PIXEL_FORMAT__BGR:
wuffs_base__store_u24le(row + (3 * ((size_t)x)), color);
wuffs_base__store_u24le__no_bounds_check(row + (3 * ((size_t)x)), color);
break;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:
wuffs_base__store_u32le(row + (4 * ((size_t)x)),
wuffs_base__nonpremul_u32_axxx(color));
wuffs_base__store_u32le__no_bounds_check(
row + (4 * ((size_t)x)), wuffs_base__nonpremul_u32_axxx(color));
break;
case WUFFS_BASE__PIXEL_FORMAT__RGB:
wuffs_base__store_u24le(row + (3 * ((size_t)x)),
wuffs_base__swap_u32_argb_abgr(color));
wuffs_base__store_u24le__no_bounds_check(
row + (3 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));
break;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:
wuffs_base__store_u32le(row + (4 * ((size_t)x)),
wuffs_base__nonpremul_u32_axxx(
wuffs_base__swap_u32_argb_abgr(color)));
wuffs_base__store_u32le__no_bounds_check(
row + (4 * ((size_t)x)), wuffs_base__nonpremul_u32_axxx(
wuffs_base__swap_u32_argb_abgr(color)));
break;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
WUFFS_BASE__FALLTHROUGH;
case WUFFS_BASE__PIXEL_FORMAT__RGBX:
wuffs_base__store_u32le(row + (4 * ((size_t)x)),
wuffs_base__swap_u32_argb_abgr(color));
wuffs_base__store_u32le__no_bounds_check(
row + (4 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));
break;
default:
@@ -273,9 +281,9 @@ wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src(
// TODO: unroll.
while (n >= 1) {
uint32_t s0 =
wuffs_base__premul_u32_axxx(wuffs_base__load_u32le(s + (0 * 4)));
wuffs_base__store_u32le(d + (0 * 4), s0);
uint32_t s0 = wuffs_base__premul_u32_axxx(
wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)));
wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);
s += 1 * 4;
d += 1 * 4;
@@ -300,9 +308,9 @@ wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over(
// TODO: unroll.
while (n >= 1) {
uint32_t d0 = wuffs_base__load_u32le(d + (0 * 4));
uint32_t s0 = wuffs_base__load_u32le(s + (0 * 4));
wuffs_base__store_u32le(
uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4));
uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4));
wuffs_base__store_u32le__no_bounds_check(
d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0));
s += 1 * 4;
@@ -349,18 +357,18 @@ wuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst,
const size_t loop_unroll_count = 4;
while (n >= loop_unroll_count) {
wuffs_base__store_u16le(
d + (0 * 2),
wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u16le(
d + (1 * 2),
wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[1] * 4)));
wuffs_base__store_u16le(
d + (2 * 2),
wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[2] * 4)));
wuffs_base__store_u16le(
d + (3 * 2),
wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[3] * 4)));
wuffs_base__store_u16le__no_bounds_check(
d + (0 * 2), wuffs_base__load_u16le__no_bounds_check(
dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u16le__no_bounds_check(
d + (1 * 2), wuffs_base__load_u16le__no_bounds_check(
dst_palette.ptr + ((size_t)s[1] * 4)));
wuffs_base__store_u16le__no_bounds_check(
d + (2 * 2), wuffs_base__load_u16le__no_bounds_check(
dst_palette.ptr + ((size_t)s[2] * 4)));
wuffs_base__store_u16le__no_bounds_check(
d + (3 * 2), wuffs_base__load_u16le__no_bounds_check(
dst_palette.ptr + ((size_t)s[3] * 4)));
s += loop_unroll_count * 1;
d += loop_unroll_count * 2;
@@ -368,9 +376,9 @@ wuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst,
}
while (n >= 1) {
wuffs_base__store_u16le(
d + (0 * 2),
wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u16le__no_bounds_check(
d + (0 * 2), wuffs_base__load_u16le__no_bounds_check(
dst_palette.ptr + ((size_t)s[0] * 4)));
s += 1 * 1;
d += 1 * 2;
@@ -402,18 +410,18 @@ wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst,
// overwrite that with the correct byte. There is always another store,
// whether a 4-byte store in this loop or a 1-byte store in the next loop.
while (n > loop_unroll_count) {
wuffs_base__store_u32le(
d + (0 * 3),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u32le(
d + (1 * 3),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4)));
wuffs_base__store_u32le(
d + (2 * 3),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4)));
wuffs_base__store_u32le(
d + (3 * 3),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[3] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (0 * 3), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (1 * 3), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[1] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (2 * 3), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[2] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (3 * 3), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[3] * 4)));
s += loop_unroll_count * 1;
d += loop_unroll_count * 3;
@@ -421,7 +429,8 @@ wuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst,
}
while (n >= 1) {
uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));
uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[0] * 4));
d[0] = (uint8_t)(s0 >> 0);
d[1] = (uint8_t)(s0 >> 8);
d[2] = (uint8_t)(s0 >> 16);
@@ -451,21 +460,25 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over(
const size_t loop_unroll_count = 4;
while (n >= loop_unroll_count) {
uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));
uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[0] * 4));
if (s0) {
wuffs_base__store_u24le(d + (0 * 4), s0);
wuffs_base__store_u24le__no_bounds_check(d + (0 * 4), s0);
}
uint32_t s1 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4));
uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[1] * 4));
if (s1) {
wuffs_base__store_u24le(d + (1 * 4), s1);
wuffs_base__store_u24le__no_bounds_check(d + (1 * 4), s1);
}
uint32_t s2 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4));
uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[2] * 4));
if (s2) {
wuffs_base__store_u24le(d + (2 * 4), s2);
wuffs_base__store_u24le__no_bounds_check(d + (2 * 4), s2);
}
uint32_t s3 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[3] * 4));
uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[3] * 4));
if (s3) {
wuffs_base__store_u24le(d + (3 * 4), s3);
wuffs_base__store_u24le__no_bounds_check(d + (3 * 4), s3);
}
s += loop_unroll_count * 1;
@@ -474,9 +487,10 @@ wuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over(
}
while (n >= 1) {
uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));
uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[0] * 4));
if (s0) {
wuffs_base__store_u24le(d + (0 * 4), s0);
wuffs_base__store_u24le__no_bounds_check(d + (0 * 4), s0);
}
s += 1 * 1;
@@ -503,18 +517,18 @@ wuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst,
const size_t loop_unroll_count = 4;
while (n >= loop_unroll_count) {
wuffs_base__store_u32le(
d + (0 * 4),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u32le(
d + (1 * 4),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4)));
wuffs_base__store_u32le(
d + (2 * 4),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4)));
wuffs_base__store_u32le(
d + (3 * 4),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[3] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (0 * 4), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (1 * 4), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[1] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (2 * 4), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[2] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (3 * 4), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[3] * 4)));
s += loop_unroll_count * 1;
d += loop_unroll_count * 4;
@@ -522,9 +536,9 @@ wuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst,
}
while (n >= 1) {
wuffs_base__store_u32le(
d + (0 * 4),
wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4)));
wuffs_base__store_u32le__no_bounds_check(
d + (0 * 4), wuffs_base__load_u32le__no_bounds_check(
dst_palette.ptr + ((size_t)s[0] * 4)));
s += 1 * 1;
d += 1 * 4;
@@ -551,21 +565,25 @@ wuffs_base__pixel_swizzler__xxxx__index_binary_alpha__src_over(
const size_t loop_unroll_count = 4;
while (n >= loop_unroll_count) {
uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));
uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[0] * 4));
if (s0) {
wuffs_base__store_u32le(d + (0 * 4), s0);
wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);
}
uint32_t s1 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4));
uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[1] * 4));
if (s1) {
wuffs_base__store_u32le(d + (1 * 4), s1);
wuffs_base__store_u32le__no_bounds_check(d + (1 * 4), s1);
}
uint32_t s2 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4));
uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[2] * 4));
if (s2) {
wuffs_base__store_u32le(d + (2 * 4), s2);
wuffs_base__store_u32le__no_bounds_check(d + (2 * 4), s2);
}
uint32_t s3 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[3] * 4));
uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[3] * 4));
if (s3) {
wuffs_base__store_u32le(d + (3 * 4), s3);
wuffs_base__store_u32le__no_bounds_check(d + (3 * 4), s3);
}
s += loop_unroll_count * 1;
@@ -574,9 +592,10 @@ wuffs_base__pixel_swizzler__xxxx__index_binary_alpha__src_over(
}
while (n >= 1) {
uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));
uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +
((size_t)s[0] * 4));
if (s0) {
wuffs_base__store_u32le(d + (0 * 4), s0);
wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);
}
s += 1 * 1;
@@ -601,8 +620,9 @@ wuffs_base__pixel_swizzler__xxxx__xxx(wuffs_base__slice_u8 dst,
// TODO: unroll.
while (n >= 1) {
wuffs_base__store_u32le(d + (0 * 4),
0xFF000000 | wuffs_base__load_u24le(s + (0 * 3)));
wuffs_base__store_u32le__no_bounds_check(
d + (0 * 4),
0xFF000000 | wuffs_base__load_u24le__no_bounds_check(s + (0 * 3)));
s += 1 * 3;
d += 1 * 4;
@@ -625,8 +645,8 @@ wuffs_base__pixel_swizzler__xxxx__y(wuffs_base__slice_u8 dst,
// TODO: unroll.
while (n >= 1) {
wuffs_base__store_u32le(d + (0 * 4),
0xFF000000 | (0x010101 * (uint32_t)s[0]));
wuffs_base__store_u32le__no_bounds_check(
d + (0 * 4), 0xFF000000 | (0x010101 * (uint32_t)s[0]));
s += 1 * 1;
d += 1 * 4;
@@ -647,11 +667,12 @@ wuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst,
size_t n = len4;
while (n--) {
uint32_t argb = wuffs_base__load_u32le(s);
uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s);
uint32_t b5 = 0x1F & (argb >> (8 - 5));
uint32_t g6 = 0x3F & (argb >> (16 - 6));
uint32_t r5 = 0x1F & (argb >> (24 - 5));
wuffs_base__store_u32le(d, (b5 << 0) | (g6 << 5) | (r5 << 11));
wuffs_base__store_u32le__no_bounds_check(
d, (b5 << 0) | (g6 << 5) | (r5 << 11));
s += 4;
d += 4;
}
+2 -2
View File
@@ -183,7 +183,7 @@ wuffs_base__io_reader__match7(uint8_t* iop_r,
uint32_t n = a & 7;
a >>= 8;
if ((io2_r - iop_r) >= 8) {
uint64_t x = wuffs_base__load_u64le(iop_r);
uint64_t x = wuffs_base__load_u64le__no_bounds_check(iop_r);
uint32_t shift = 8 * (8 - n);
return ((a << shift) == (x << shift)) ? 0 : 2;
}
@@ -251,7 +251,7 @@ wuffs_base__io_writer__set(wuffs_base__io_buffer* b,
return b;
}
// ---------------- I/O (Utility)
// ---------------- I/O (Utility)
#define wuffs_base__utility__empty_io_reader wuffs_base__empty_io_reader
#define wuffs_base__utility__empty_io_writer wuffs_base__empty_io_writer
+3 -3
View File
@@ -214,7 +214,7 @@ func (g *gen) writeBuiltinIOReader(b *buffer, recv *a.Expr, method t.ID, args []
if p.size != p.n {
b.printf("((uint%d_t)(", p.size)
}
b.printf("wuffs_base__load_u%d%ce(%s%s)", p.n, p.endianness, iopPrefix, name)
b.printf("wuffs_base__load_u%d%ce__no_bounds_check(%s%s)", p.n, p.endianness, iopPrefix, name)
if p.size != p.n {
b.writes("))")
}
@@ -308,7 +308,7 @@ func (g *gen) writeBuiltinIOWriter(b *buffer, recv *a.Expr, method t.ID, args []
// final part is a function call (to a static inline function)
// instead of a struct literal, to avoid a "expression result
// unused" compiler error.
b.printf("(wuffs_base__store_u%d%ce(%s%s,", p.n, p.endianness, iopPrefix, name)
b.printf("(wuffs_base__store_u%d%ce__no_bounds_check(%s%s,", p.n, p.endianness, iopPrefix, name)
if err := g.writeExpr(b, args[0].AsArg().Value(), depth); err != nil {
return err
}
@@ -746,7 +746,7 @@ func (g *gen) writeReadUxxAsUyy(b *buffer, n *a.Expr, preName string, xx uint8,
if xx != yy {
b.printf("((uint%d_t)(", yy)
}
b.printf("wuffs_base__load_u%d%ce(iop_a_src)", xx, endianness)
b.printf("wuffs_base__load_u%d%ce__no_bounds_check(iop_a_src)", xx, endianness)
if xx != yy {
b.writes("))")
}
+37 -35
View File
@@ -30,27 +30,28 @@ const baseImageImplC = "" +
"// ---------------- Images\n\nconst uint32_t wuffs_base__pixel_format__bits_per_channel[16] = {\n 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,\n 0x08, 0x0A, 0x0C, 0x10, 0x18, 0x20, 0x30, 0x40,\n};\n\nstatic inline uint32_t //\nwuffs_base__swap_u32_argb_abgr(uint32_t u) {\n uint32_t o = u & 0xFF00FF00;\n uint32_t r = u & 0x00FF0000;\n uint32_t b = u & 0x000000FF;\n return o | (r >> 16) | (b << 16);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_premul_nonpremul_u32_axxx(uint32_t dst_premul,\n uint32_t src_nonpremul) {\n // Convert from 8-bit color to 16-bit color.\n uint32_t sa = 0x101 * (0xFF & (src_nonpremul >> 24));\n uint32_t sr = 0x101 * (0xFF & (src_nonpremul >> 16));\n uint32_t sg = 0x101 * (0xFF & (src_nonpremul >> 8));\n uint32_t sb = 0x101 * (0xFF & (src_nonpremul >> 0));\n uint32_t da = 0x101 * (0xFF & (dst_premul >> 24));\n uint32_t dr = 0x101 * (0xFF & (dst_premul >> 16));\n uint32_t dg = 0x101 * (0xFF & (dst_premul >> 8));\n uint32_t db = 0x101" +
" * (0xFF & (dst_premul >> 0));\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul).\n da = sa + ((da * ia) / 0xFFFF);\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Convert from 16-bit color to 8-bit color and combine the components.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint32_t //\nwuffs_base__premul_u32_axxx(uint32_t nonpremul) {\n // Multiplying by 0x101 (twice, once for alpha and once for color) converts\n // from 8-bit to 16-bit color. Shifting right by 8 undoes that.\n //\n // Working in the higher bit depth can produce slightly different (and\n // arguably slightly more accurate) results. For example, given 8-bit blue\n // and alpha of 0x80 and 0x81:\n //\n // - ((0x80 * 0x81 ) / 0xFF ) = 0x40 = 0x40\n // - ((0x8080 * 0" +
"x8181) / 0xFFFF) >> 8 = 0x4101 >> 8 = 0x41\n uint32_t a = 0xFF & (nonpremul >> 24);\n uint32_t a16 = a * (0x101 * 0x101);\n\n uint32_t r = 0xFF & (nonpremul >> 16);\n r = ((r * a16) / 0xFFFF) >> 8;\n uint32_t g = 0xFF & (nonpremul >> 8);\n g = ((g * a16) / 0xFFFF) >> 8;\n uint32_t b = 0xFF & (nonpremul >> 0);\n b = ((b * a16) / 0xFFFF) >> 8;\n\n return (a << 24) | (r << 16) | (g << 8) | (b << 0);\n}\n\nstatic inline uint32_t //\nwuffs_base__nonpremul_u32_axxx(uint32_t premul) {\n uint32_t a = 0xFF & (premul >> 24);\n if (a == 0xFF) {\n return premul;\n } else if (a == 0) {\n return 0;\n }\n uint32_t a16 = a * 0x101;\n\n uint32_t r = 0xFF & (premul >> 16);\n r = ((r * (0x101 * 0xFFFF)) / a16) >> 8;\n uint32_t g = 0xFF & (premul >> 8);\n g = ((g * (0x101 * 0xFFFF)) / a16) >> 8;\n uint32_t b = 0xFF & (premul >> 0);\n b = ((b * (0x101 * 0xFFFF)) / a16) >> 8;\n\n return (a << 24) | (r << 16) | (g << 8) | (b << 0);\n}\n\nwuffs_base__color_u32_argb_premul //\nwuffs_base__pixel_buffer__color_u32_at(const wuffs_base__pixel_" +
"buffer* b,\n uint32_t x,\n uint32_t y) {\n if (!b || (x >= b->pixcfg.private_impl.width) ||\n (y >= b->pixcfg.private_impl.height)) {\n return 0;\n }\n\n if (wuffs_base__pixel_format__is_planar(&b->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return 0;\n }\n\n size_t stride = b->private_impl.planes[0].stride;\n uint8_t* row = b->private_impl.planes[0].ptr + (stride * ((size_t)y));\n\n switch (b->pixcfg.private_impl.pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n return wuffs_base__load_u32le(row + (4 * ((size_t)x)));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY: {\n uint8_t* palette = b->private_impl.planes[3].ptr;\n return wuffs_base__load_u32le(palette + (4 * ((size_t)row[x])));\n }\n\n // Common" +
" formats above. Rarer formats below.\n\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n return 0xFF000000 | (0x00010101 * ((uint32_t)(row[x])));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL: {\n uint8_t* palette = b->private_impl.planes[3].ptr;\n return wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le(palette + (4 * ((size_t)row[x]))));\n }\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565: {\n uint16_t bgr = wuffs_base__load_u16le(row + (2 * ((size_t)x)));\n uint32_t b5 = 0x1F & (bgr >> 0);\n uint32_t b = (b5 << 3) | (b5 >> 2);\n uint32_t g6 = 0x3F & (bgr >> 5);\n uint32_t g = (g6 << 2) | (g6 >> 4);\n uint32_t r5 = 0x1F & (bgr >> 11);\n uint32_t r = (r5 << 3) | (r5 >> 2);\n return 0xFF000000 | (r << 16) | (g << 8) | (b << 0);\n }\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return 0xFF000000 | wuffs_base__load_u24le(row + (3 * ((size_t)x)));\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n return 0xFF000000 | wuffs_base__load_u32le(row + (4 * ((size_" +
"t)x)));\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n return wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le(row + (4 * ((size_t)x))));\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 | wuffs_base__load_u24le(row + (3 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 | wuffs_base__load_u32le(row + (4 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n return wuffs_base__swap_u32_argb_abgr(wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le(row + (4 * ((size_t)x)))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:\n return wuffs_base__swap_u32_argb_abgr(\n wuffs_base__load_u32le(row + (4 * ((size_t)x))));\n\n default:\n // TODO: support more formats.\n break;\n }\n\n return 0;\n}\n\nwuffs_base__status //\nwuffs_base__pixel_buf" +
"fer__set_color_u32_at(\n wuffs_base__pixel_buffer* b,\n uint32_t x,\n uint32_t y,\n wuffs_base__color_u32_argb_premul color) {\n if (!b) {\n return wuffs_base__make_status(wuffs_base__error__bad_receiver);\n }\n if ((x >= b->pixcfg.private_impl.width) ||\n (y >= b->pixcfg.private_impl.height)) {\n return wuffs_base__make_status(wuffs_base__error__bad_argument);\n }\n\n if (wuffs_base__pixel_format__is_planar(&b->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n size_t stride = b->private_impl.planes[0].stride;\n uint8_t* row = b->private_impl.planes[0].ptr + (stride * ((size_t)y));\n\n switch (b->pixcfg.private_impl.pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n wuffs_base__store_u32le(row + (4 * ((size_t)x)), color);\n break;\n\n // Common formats above. Rarer formats below.\n\n case WUFFS_BASE__" +
"PIXEL_FORMAT__BGR_565: {\n uint32_t b5 = 0x1F & (color >> (8 - 5));\n uint32_t g6 = 0x3F & (color >> (16 - 6));\n uint32_t r5 = 0x1F & (color >> (24 - 5));\n uint32_t bgr565 = (b5 << 0) | (g6 << 5) | (r5 << 11);\n wuffs_base__store_u16le(row + (2 * ((size_t)x)), (uint16_t)bgr565);\n break;\n }\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n wuffs_base__store_u24le(row + (3 * ((size_t)x)), color);\n break;\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n wuffs_base__store_u32le(row + (4 * ((size_t)x)),\n wuffs_base__nonpremul_u32_axxx(color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n wuffs_base__store_u24le(row + (3 * ((size_t)x)),\n wuffs_base__swap_u32_argb_abgr(color));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n wuffs_base__store_u32le(row + (4 * ((size_t)x)),\n wuffs_base__nonpremul_u32_axxx(\n wuffs_base__swap_" +
"u32_argb_abgr(color)));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n wuffs_base__store_u32le(row + (4 * ((size_t)x)),\n wuffs_base__swap_u32_argb_abgr(color));\n break;\n\n default:\n // TODO: support more formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n return wuffs_base__make_status(NULL);\n}\n\n" +
"buffer* b,\n uint32_t x,\n uint32_t y) {\n if (!b || (x >= b->pixcfg.private_impl.width) ||\n (y >= b->pixcfg.private_impl.height)) {\n return 0;\n }\n\n if (wuffs_base__pixel_format__is_planar(&b->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return 0;\n }\n\n size_t stride = b->private_impl.planes[0].stride;\n uint8_t* row = b->private_impl.planes[0].ptr + (stride * ((size_t)y));\n\n switch (b->pixcfg.private_impl.pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n return wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY: {\n uint8_t* palette = b->private_impl.planes[3].ptr;\n return wuffs_base__load_u32le__no_bounds_check(palette +\n " +
" (4 * ((size_t)row[x])));\n }\n\n // Common formats above. Rarer formats below.\n\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n return 0xFF000000 | (0x00010101 * ((uint32_t)(row[x])));\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL: {\n uint8_t* palette = b->private_impl.planes[3].ptr;\n return wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(palette +\n (4 * ((size_t)row[x]))));\n }\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565: {\n uint16_t bgr =\n wuffs_base__load_u16le__no_bounds_check(row + (2 * ((size_t)x)));\n uint32_t b5 = 0x1F & (bgr >> 0);\n uint32_t b = (b5 << 3) | (b5 >> 2);\n uint32_t g6 = 0x3F & (bgr >> 5);\n uint32_t g = (g6 << 2) | (g6 >> 4);\n uint32_t r5 = 0x1F & (bgr >> 11);\n uint32_t r = (r5 << 3) | (r5 >> 2);\n return 0xFF000000 | (r << 16) | (g << 8) | (b << 0);\n }\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n " +
" return 0xFF000000 |\n wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x)));\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n return 0xFF000000 |\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)));\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n return wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 |\n wuffs_base__load_u24le__no_bounds_check(row + (3 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n return wuffs_base__swap_u32_argb_abgr(\n 0xFF000000 |\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n return wuffs_base__swap_u32_argb_abgr(wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x)))));\n case WUFFS_BASE__PIXEL_FORM" +
"AT__RGBA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:\n return wuffs_base__swap_u32_argb_abgr(\n wuffs_base__load_u32le__no_bounds_check(row + (4 * ((size_t)x))));\n\n default:\n // TODO: support more formats.\n break;\n }\n\n return 0;\n}\n\nwuffs_base__status //\nwuffs_base__pixel_buffer__set_color_u32_at(\n wuffs_base__pixel_buffer* b,\n uint32_t x,\n uint32_t y,\n wuffs_base__color_u32_argb_premul color) {\n if (!b) {\n return wuffs_base__make_status(wuffs_base__error__bad_receiver);\n }\n if ((x >= b->pixcfg.private_impl.width) ||\n (y >= b->pixcfg.private_impl.height)) {\n return wuffs_base__make_status(wuffs_base__error__bad_argument);\n }\n\n if (wuffs_base__pixel_format__is_planar(&b->pixcfg.private_impl.pixfmt)) {\n // TODO: support planar formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n size_t stride = b->private_impl.planes[0].stride;\n uint8_t* row = b->private_impl.planes[0].ptr " +
"+ (stride * ((size_t)y));\n\n switch (b->pixcfg.private_impl.pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n wuffs_base__store_u32le__no_bounds_check(row + (4 * ((size_t)x)), color);\n break;\n\n // Common formats above. Rarer formats below.\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565: {\n uint32_t b5 = 0x1F & (color >> (8 - 5));\n uint32_t g6 = 0x3F & (color >> (16 - 6));\n uint32_t r5 = 0x1F & (color >> (24 - 5));\n uint32_t bgr565 = (b5 << 0) | (g6 << 5) | (r5 << 11);\n wuffs_base__store_u16le__no_bounds_check(row + (2 * ((size_t)x)),\n (uint16_t)bgr565);\n break;\n }\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n wuffs_base__store_u24le__no_bounds_check(row + (3 * ((size_t)x)), color);\n break;\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_bas" +
"e__nonpremul_u32_axxx(color));\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n wuffs_base__store_u24le__no_bounds_check(\n row + (3 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__nonpremul_u32_axxx(\n wuffs_base__swap_u32_argb_abgr(color)));\n break;\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n WUFFS_BASE__FALLTHROUGH;\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n wuffs_base__store_u32le__no_bounds_check(\n row + (4 * ((size_t)x)), wuffs_base__swap_u32_argb_abgr(color));\n break;\n\n default:\n // TODO: support more formats.\n return wuffs_base__make_status(wuffs_base__error__unsupported_option);\n }\n\n return wuffs_base__make_status(NULL);\n}\n\n" +
"" +
"// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len4 < src_len4 ? dst_len4 : src_len4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 =\n wuffs_base__premul_u32_axxx(wuffs_base__load_u32le(s + (0 * 4)));\n wuffs_base__store_u32le(d + (0 * 4), s0);\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len4 < src_len4 ? dst_len4 : src_len4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unrol" +
"l.\n\n while (n >= 1) {\n uint32_t d0 = wuffs_base__load_u32le(d + (0 * 4));\n uint32_t s0 = wuffs_base__load_u32le(s + (0 * 4));\n wuffs_base__store_u32le(\n d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_1_1(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n return wuffs_base__slice_u8__copy_from_slice(dst, src);\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len4 < src_len4 ? dst_len4 : src_len4;\n if (len > 0) {\n memmove(dst.ptr, src.ptr, len * 4);\n }\n return len;\n}\n\nstatic uint64_t" +
" //\nwuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst.len / 2;\n size_t len = dst_len2 < src.len ? dst_len2 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__store_u16le(\n d + (0 * 2),\n wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u16le(\n d + (1 * 2),\n wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u16le(\n d + (2 * 2),\n wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u16le(\n d + (3 * 2),\n wuffs_base__load_u16le(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(\n d + (0 * 2),\n wuffs_base__load_u16le(dst_palette.ptr + ((size_t)s[0] * 4)));\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst.len / 3;\n size_t len = dst_len3 < src.len ? dst_len3 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n // The comparison in the while condition is \">\", not \">=\", because with \">=\",\n // the last 4-byte store could write past the end of the dst slice.\n //\n // Each 4-byte store writes one too many bytes, but a subsequent store will\n // overwrite that with the correct " +
"byte. There is always another store,\n // whether a 4-byte store in this loop or a 1-byte store in the next loop.\n while (n > loop_unroll_count) {\n wuffs_base__store_u32le(\n d + (0 * 3),\n wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u32le(\n d + (1 * 3),\n wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u32le(\n d + (2 * 3),\n wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u32le(\n d + (3 * 3),\n wuffs_base__load_u32le(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(dst_palette.ptr + ((size_t)s[0] * 4));\n d[0] = (uint8_t)(s0 >> 0);\n d[1] = (uint8_t)(s0 >> 8);\n d[2] = (uint8_t)(s0 >> 16);\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pi" +
"xel_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(dst_palette.ptr + ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u24le(d + (0 * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u24le(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u24le(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[3] * 4));\n if (s3) {\n wuffs_base__store_u24le(d + (3 * 4), s3);\n }\n\n " +
" s += loop_unroll_count * 1;\n d += loop_unroll_count * 3;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u24le(d + (0 * 4), s0);\n }\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len4 = dst.len / 4;\n size_t len = dst_len4 < src.len ? dst_len4 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__store_u32le(\n d + (0 * 4),\n wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u32le(\n d + (1 * 4)," +
"\n wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u32le(\n d + (2 * 4),\n wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u32le(\n d + (3 * 4),\n wuffs_base__load_u32le(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(\n d + (0 * 4),\n wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4)));\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxxx__index_binary_alpha__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len4 = dst.len / 4;\n size_t len = dst_len4 < src.len ? dst_len4 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t" +
" loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le(d + (0 * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u32le(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u32le(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[3] * 4));\n if (s3) {\n wuffs_base__store_u32le(d + (3 * 4), s3);\n }\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 4;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le(dst_palette.ptr + ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le(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_sw" +
"izzler__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(d + (0 * 4),\n 0xFF000000 | wuffs_base__load_u24le(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(d + (0 * 4),\n 0xFF000000 | (0x010101 * (uint32_t)s[0]));\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" +
"// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len4 < src_len4 ? dst_len4 : src_len4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__premul_u32_axxx(\n wuffs_base__load_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len4 < src_len4 ? dst_len4 : src_len4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n" +
" size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t d0 = wuffs_base__load_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_1_1(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n return wuffs_base__slice_u8__copy_from_slice(dst, src);\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__copy_4_4(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t src_len4 = src.len / 4;\n size_t len = dst_len4 < src_len4 ? dst_len4 : src_len4;\n if (l" +
"en > 0) {\n memmove(dst.ptr, src.ptr, len * 4);\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst.len / 2;\n size_t len = dst_len2 < src.len ? dst_len2 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__store_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (1 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (2 * 2), wuffs_base__load_u16le" +
"__no_bounds_check(\n dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_base__store_u16le__no_bounds_check(\n d + (3 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 2;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n wuffs_base__store_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__load_u16le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst.len / 3;\n size_t len = dst_len3 < src.len ? dst_len3 : src.len;\n uint8_t* d = dst.ptr;\n uint" +
"8_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 will\n // overwrite that with the correct byte. There is always another store,\n // whether a 4-byte store in this loop or a 1-byte store in the next loop.\n while (n > loop_unroll_count) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[0] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (1 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[1] * 4)));\n wuffs_base__store_u32le__no_bounds_check(\n d + (2 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[2] * 4)));\n wuffs_b" +
"ase__store_u32le__no_bounds_check(\n d + (3 * 3), wuffs_base__load_u32le__no_bounds_check(\n dst_palette.ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 3;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n d[0] = (uint8_t)(s0 >> 0);\n d[1] = (uint8_t)(s0 >> 8);\n d[2] = (uint8_t)(s0 >> 16);\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxx__index_binary_alpha__src_over(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len3 = dst.len / 3;\n size_t len = dst_len3 < src.len ? dst_len3 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n" +
" while (n >= loop_unroll_count) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u24le__no_bounds_check(d + (0 * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u24le__no_bounds_check(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[2] * 4));\n if (s2) {\n wuffs_base__store_u24le__no_bounds_check(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[3] * 4));\n if (s3) {\n wuffs_base__store_u24le__no_bounds_check(d + (3 * 4), s3);\n }\n\n s += loop_unroll_count *" +
" 1;\n d += loop_unroll_count * 3;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u24le__no_bounds_check(d + (0 * 4), s0);\n }\n\n s += 1 * 1;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxxx__index__src(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len4 = dst.len / 4;\n size_t len = dst_len4 < src.len ? dst_len4 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__load_u32le__no_bounds_chec" +
"k(\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_ov" +
"er(\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (dst_palette.len != 1024) {\n return 0;\n }\n size_t dst_len4 = dst.len / 4;\n size_t len = dst_len4 < src.len ? dst_len4 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);\n }\n uint32_t s1 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[1] * 4));\n if (s1) {\n wuffs_base__store_u32le__no_bounds_check(d + (1 * 4), s1);\n }\n uint32_t s2 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[2] * 4));\n " +
"if (s2) {\n wuffs_base__store_u32le__no_bounds_check(d + (2 * 4), s2);\n }\n uint32_t s3 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[3] * 4));\n if (s3) {\n wuffs_base__store_u32le__no_bounds_check(d + (3 * 4), s3);\n }\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 4;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__load_u32le__no_bounds_check(dst_palette.ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__store_u32le__no_bounds_check(d + (0 * 4), s0);\n }\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxxx__xxx(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t src_len3 = src." +
"len / 3;\n size_t len = dst_len4 < src_len3 ? dst_len4 : src_len3;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4),\n 0xFF000000 | wuffs_base__load_u24le__no_bounds_check(s + (0 * 3)));\n\n s += 1 * 3;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__xxxx__y(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n size_t dst_len4 = dst.len / 4;\n size_t len = dst_len4 < src.len ? dst_len4 : src.len;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__store_u32le__no_bounds_check(\n d + (0 * 4), 0xFF000000 | (0x010101 * (uint32_t)s[0]));\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" +
"" +
"// --------\n\nstatic 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(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(d, (b5 << 0) | (g6 << 5) | (r5 << 11));\n s += 4;\n d += 4;\n }\n return len4 * 4;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 src) {\n size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n\n size_t n = len4;\n while (n--) {\n uint8_t b0 = s[0];\n uint8_t b1 = s[1];\n 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\n" +
"// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__squash_bgr_565_888(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 src) {\n size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n\n size_t n = len4;\n while (n--) {\n uint32_t argb = wuffs_base__load_u32le__no_bounds_check(s);\n uint32_t b5 = 0x1F & (argb >> (8 - 5));\n uint32_t g6 = 0x3F & (argb >> (16 - 6));\n uint32_t r5 = 0x1F & (argb >> (24 - 5));\n wuffs_base__store_u32le__no_bounds_check(\n d, (b5 << 0) | (g6 << 5) | (r5 << 11));\n s += 4;\n d += 4;\n }\n return len4 * 4;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__swap_rgbx_bgrx(wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 src) {\n size_t len4 = (dst.len < src.len ? dst.len : src.len) / 4;\n uint8_t* d = dst.ptr;\n uint8_t* s = src.ptr;\n\n size_t n = len4;\n while (n--) {\n uint8_t b0 = s[0];\n uint8_t b1 = s[1];\n u" +
"int8_t b2 = s[2];\n uint8_t b3 = s[3];\n d[0] = b2;\n d[1] = b1;\n d[2] = b0;\n d[3] = b3;\n s += 4;\n d += 4;\n }\n return len4 * 4;\n}\n\n" +
"" +
"// --------\n\nstatic 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__BGRX:\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__RGBX:\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 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 " +
@@ -82,19 +83,9 @@ const baseStrConvImplC = "" +
const baseCorePrivateH = "" +
"static inline wuffs_base__empty_struct //\nwuffs_base__ignore_status(wuffs_base__status z) {\n return wuffs_base__make_empty_struct();\n}\n\n// WUFFS_BASE__MAGIC is a magic number to check that initializers are called.\n// It's not foolproof, given C doesn't automatically zero memory before use,\n// but it should catch 99.99% of cases.\n//\n// Its (non-zero) value is arbitrary, based on md5sum(\"wuffs\").\n#define WUFFS_BASE__MAGIC ((uint32_t)0x3CCB6C71)\n\n// WUFFS_BASE__DISABLED is a magic number to indicate that a non-recoverable\n// error was previously encountered.\n//\n// Its (non-zero) value is arbitrary, based on md5sum(\"disabled\").\n#define WUFFS_BASE__DISABLED ((uint32_t)0x075AE3D2)\n\n// Denote intentional fallthroughs for -Wimplicit-fallthrough.\n//\n// The order matters here. Clang also defines \"__GNUC__\".\n#if defined(__clang__) && defined(__cplusplus) && (__cplusplus >= 201103L)\n#define WUFFS_BASE__FALLTHROUGH [[clang::fallthrough]]\n#elif !defined(__clang__) && defined(__GNUC__) && (__GNUC__ >= 7)\n#define WUFFS_BAS" +
"E__FALLTHROUGH __attribute__((fallthrough))\n#else\n#define WUFFS_BASE__FALLTHROUGH\n#endif\n\n// Use switch cases for coroutine suspension points, similar to the technique\n// in https://www.chiark.greenend.org.uk/~sgtatham/coroutines.html\n//\n// We use trivial macros instead of an explicit assignment and case statement\n// so that clang-format doesn't get confused by the unusual \"case\"s.\n#define WUFFS_BASE__COROUTINE_SUSPENSION_POINT_0 case 0:;\n#define WUFFS_BASE__COROUTINE_SUSPENSION_POINT(n) \\\n coro_susp_point = n; \\\n WUFFS_BASE__FALLTHROUGH; \\\n case n:;\n\n#define WUFFS_BASE__COROUTINE_SUSPENSION_POINT_MAYBE_SUSPEND(n) \\\n if (!status.repr) { \\\n goto ok; \\\n } else if (*status.repr != '$') { \\\n goto exit; \\\n } \\\n coro_susp_point" +
" = n; \\\n goto suspend; \\\n case n:;\n\n// Clang also defines \"__GNUC__\".\n#if defined(__GNUC__)\n#define WUFFS_BASE__LIKELY(expr) (__builtin_expect(!!(expr), 1))\n#define WUFFS_BASE__UNLIKELY(expr) (__builtin_expect(!!(expr), 0))\n#else\n#define WUFFS_BASE__LIKELY(expr) (expr)\n#define WUFFS_BASE__UNLIKELY(expr) (expr)\n#endif\n\n// The helpers below are functions, instead of macros, because their arguments\n// can be an expression that we shouldn't evaluate more than once.\n//\n// They are static, so that linking multiple wuffs .o files won't complain about\n// duplicate function definitions.\n//\n// They are explicitly marked inline, even if modern compilers don't use the\n// inline attribute to guide optimizations such as inlining, to avoid the\n// -Wunused-function warning, and we like to compile with -Wall -Werror.\n\n" +
" = n; \\\n goto suspend; \\\n case n:;\n\n// Clang also defines \"__GNUC__\".\n#if defined(__GNUC__)\n#define WUFFS_BASE__LIKELY(expr) (__builtin_expect(!!(expr), 1))\n#define WUFFS_BASE__UNLIKELY(expr) (__builtin_expect(!!(expr), 0))\n#else\n#define WUFFS_BASE__LIKELY(expr) (expr)\n#define WUFFS_BASE__UNLIKELY(expr) (expr)\n#endif\n\n" +
"" +
"// ---------------- Numeric Types\n\nstatic inline uint8_t //\nwuffs_base__load_u8be(uint8_t* p) {\n return p[0];\n}\n\nstatic inline uint16_t //\nwuffs_base__load_u16be(uint8_t* p) {\n return (uint16_t)(((uint16_t)(p[0]) << 8) | ((uint16_t)(p[1]) << 0));\n}\n\nstatic inline uint16_t //\nwuffs_base__load_u16le(uint8_t* p) {\n return (uint16_t)(((uint16_t)(p[0]) << 0) | ((uint16_t)(p[1]) << 8));\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u24be(uint8_t* p) {\n return ((uint32_t)(p[0]) << 16) | ((uint32_t)(p[1]) << 8) |\n ((uint32_t)(p[2]) << 0);\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u24le(uint8_t* p) {\n return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]) << 8) |\n ((uint32_t)(p[2]) << 16);\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u32be(uint8_t* p) {\n return ((uint32_t)(p[0]) << 24) | ((uint32_t)(p[1]) << 16) |\n ((uint32_t)(p[2]) << 8) | ((uint32_t)(p[3]) << 0);\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u32le(uint8_t* p) {\n return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]" +
") << 8) |\n ((uint32_t)(p[2]) << 16) | ((uint32_t)(p[3]) << 24);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u40be(uint8_t* p) {\n return ((uint64_t)(p[0]) << 32) | ((uint64_t)(p[1]) << 24) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 8) |\n ((uint64_t)(p[4]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u40le(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u48be(uint8_t* p) {\n return ((uint64_t)(p[0]) << 40) | ((uint64_t)(p[1]) << 32) |\n ((uint64_t)(p[2]) << 24) | ((uint64_t)(p[3]) << 16) |\n ((uint64_t)(p[4]) << 8) | ((uint64_t)(p[5]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u48le(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) <<" +
" 40);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u56be(uint8_t* p) {\n return ((uint64_t)(p[0]) << 48) | ((uint64_t)(p[1]) << 40) |\n ((uint64_t)(p[2]) << 32) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 16) | ((uint64_t)(p[5]) << 8) |\n ((uint64_t)(p[6]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u56le(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |\n ((uint64_t)(p[6]) << 48);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u64be(uint8_t* p) {\n return ((uint64_t)(p[0]) << 56) | ((uint64_t)(p[1]) << 48) |\n ((uint64_t)(p[2]) << 40) | ((uint64_t)(p[3]) << 32) |\n ((uint64_t)(p[4]) << 24) | ((uint64_t)(p[5]) << 16) |\n ((uint64_t)(p[6]) << 8) | ((uint64_t)(p[7]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u64le(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8" +
") |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |\n ((uint64_t)(p[6]) << 48) | ((uint64_t)(p[7]) << 56);\n}\n\n" +
"" +
"// --------\n\nstatic inline void //\nwuffs_base__store_u8be(uint8_t* p, uint8_t x) {\n p[0] = x;\n}\n\nstatic inline void //\nwuffs_base__store_u16be(uint8_t* p, uint16_t x) {\n p[0] = (uint8_t)(x >> 8);\n p[1] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u16le(uint8_t* p, uint16_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n}\n\nstatic inline void //\nwuffs_base__store_u24be(uint8_t* p, uint32_t x) {\n p[0] = (uint8_t)(x >> 16);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u24le(uint8_t* p, uint32_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n}\n\nstatic inline void //\nwuffs_base__store_u32be(uint8_t* p, uint32_t x) {\n p[0] = (uint8_t)(x >> 24);\n p[1] = (uint8_t)(x >> 16);\n p[2] = (uint8_t)(x >> 8);\n p[3] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u32le(uint8_t* p, uint32_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x" +
" >> 16);\n p[3] = (uint8_t)(x >> 24);\n}\n\nstatic inline void //\nwuffs_base__store_u40be(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 32);\n p[1] = (uint8_t)(x >> 24);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 8);\n p[4] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u40le(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32);\n}\n\nstatic inline void //\nwuffs_base__store_u48be(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 40);\n p[1] = (uint8_t)(x >> 32);\n p[2] = (uint8_t)(x >> 24);\n p[3] = (uint8_t)(x >> 16);\n p[4] = (uint8_t)(x >> 8);\n p[5] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u48le(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32);\n p[5] = (uint8_t)(x >> 40);\n}\n\nstatic inline void //\nwuffs_base__store_u56be(ui" +
"nt8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 48);\n p[1] = (uint8_t)(x >> 40);\n p[2] = (uint8_t)(x >> 32);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 16);\n p[5] = (uint8_t)(x >> 8);\n p[6] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u56le(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32);\n p[5] = (uint8_t)(x >> 40);\n p[6] = (uint8_t)(x >> 48);\n}\n\nstatic inline void //\nwuffs_base__store_u64be(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 56);\n p[1] = (uint8_t)(x >> 48);\n p[2] = (uint8_t)(x >> 40);\n p[3] = (uint8_t)(x >> 32);\n p[4] = (uint8_t)(x >> 24);\n p[5] = (uint8_t)(x >> 16);\n p[6] = (uint8_t)(x >> 8);\n p[7] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u64le(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32" +
");\n p[5] = (uint8_t)(x >> 40);\n p[6] = (uint8_t)(x >> 48);\n p[7] = (uint8_t)(x >> 56);\n}\n\n" +
"" +
"// --------\n\nextern const uint8_t wuffs_base__low_bits_mask__u8[9];\nextern const uint16_t wuffs_base__low_bits_mask__u16[17];\nextern const uint32_t wuffs_base__low_bits_mask__u32[33];\nextern const uint64_t wuffs_base__low_bits_mask__u64[65];\n\n#define WUFFS_BASE__LOW_BITS_MASK__U8(n) (wuffs_base__low_bits_mask__u8[n])\n#define WUFFS_BASE__LOW_BITS_MASK__U16(n) (wuffs_base__low_bits_mask__u16[n])\n#define WUFFS_BASE__LOW_BITS_MASK__U32(n) (wuffs_base__low_bits_mask__u32[n])\n#define WUFFS_BASE__LOW_BITS_MASK__U64(n) (wuffs_base__low_bits_mask__u64[n])\n\n" +
"// ---------------- Numeric Types\n\nextern const uint8_t wuffs_base__low_bits_mask__u8[9];\nextern const uint16_t wuffs_base__low_bits_mask__u16[17];\nextern const uint32_t wuffs_base__low_bits_mask__u32[33];\nextern const uint64_t wuffs_base__low_bits_mask__u64[65];\n\n#define WUFFS_BASE__LOW_BITS_MASK__U8(n) (wuffs_base__low_bits_mask__u8[n])\n#define WUFFS_BASE__LOW_BITS_MASK__U16(n) (wuffs_base__low_bits_mask__u16[n])\n#define WUFFS_BASE__LOW_BITS_MASK__U32(n) (wuffs_base__low_bits_mask__u32[n])\n#define WUFFS_BASE__LOW_BITS_MASK__U64(n) (wuffs_base__low_bits_mask__u64[n])\n\n" +
"" +
"// --------\n\nstatic inline void //\nwuffs_base__u8__sat_add_indirect(uint8_t* x, uint8_t y) {\n *x = wuffs_base__u8__sat_add(*x, y);\n}\n\nstatic inline void //\nwuffs_base__u8__sat_sub_indirect(uint8_t* x, uint8_t y) {\n *x = wuffs_base__u8__sat_sub(*x, y);\n}\n\nstatic inline void //\nwuffs_base__u16__sat_add_indirect(uint16_t* x, uint16_t y) {\n *x = wuffs_base__u16__sat_add(*x, y);\n}\n\nstatic inline void //\nwuffs_base__u16__sat_sub_indirect(uint16_t* x, uint16_t y) {\n *x = wuffs_base__u16__sat_sub(*x, y);\n}\n\nstatic inline void //\nwuffs_base__u32__sat_add_indirect(uint32_t* x, uint32_t y) {\n *x = wuffs_base__u32__sat_add(*x, y);\n}\n\nstatic inline void //\nwuffs_base__u32__sat_sub_indirect(uint32_t* x, uint32_t y) {\n *x = wuffs_base__u32__sat_sub(*x, y);\n}\n\nstatic inline void //\nwuffs_base__u64__sat_add_indirect(uint64_t* x, uint64_t y) {\n *x = wuffs_base__u64__sat_add(*x, y);\n}\n\nstatic inline void //\nwuffs_base__u64__sat_sub_indirect(uint64_t* x, uint64_t y) {\n *x = wuffs_base__u64__sat_sub(*x, y);\n}\n\n" +
"" +
@@ -126,10 +117,21 @@ const baseCorePublicH = "" +
"" +
"// --------\n\n// Flicks are a unit of time. One flick (frame-tick) is 1 / 705_600_000 of a\n// second. See https://github.com/OculusVR/Flicks\ntypedef int64_t wuffs_base__flicks;\n\n#define WUFFS_BASE__FLICKS_PER_SECOND ((uint64_t)705600000)\n#define WUFFS_BASE__FLICKS_PER_MILLISECOND ((uint64_t)705600)\n\n" +
"" +
"// ---------------- Numeric Types\n\nstatic inline uint8_t //\nwuffs_base__u8__min(uint8_t x, uint8_t y) {\n return x < y ? x : y;\n}\n\nstatic inline uint8_t //\nwuffs_base__u8__max(uint8_t x, uint8_t y) {\n return x > y ? x : y;\n}\n\nstatic inline uint16_t //\nwuffs_base__u16__min(uint16_t x, uint16_t y) {\n return x < y ? x : y;\n}\n\nstatic inline uint16_t //\nwuffs_base__u16__max(uint16_t x, uint16_t y) {\n return x > y ? x : y;\n}\n\nstatic inline uint32_t //\nwuffs_base__u32__min(uint32_t x, uint32_t y) {\n return x < y ? x : y;\n}\n\nstatic inline uint32_t //\nwuffs_base__u32__max(uint32_t x, uint32_t y) {\n return x > y ? x : y;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__min(uint64_t x, uint64_t y) {\n return x < y ? x : y;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__max(uint64_t x, uint64_t y) {\n return x > y ? x : y;\n}\n\n" +
"// ---------------- Numeric Types\n\n// The helpers below are functions, instead of macros, because their arguments\n// can be an expression that we shouldn't evaluate more than once.\n//\n// They are static, so that linking multiple wuffs .o files won't complain about\n// duplicate function definitions.\n//\n// They are explicitly marked inline, even if modern compilers don't use the\n// inline attribute to guide optimizations such as inlining, to avoid the\n// -Wunused-function warning, and we like to compile with -Wall -Werror.\n\nstatic inline uint8_t //\nwuffs_base__u8__min(uint8_t x, uint8_t y) {\n return x < y ? x : y;\n}\n\nstatic inline uint8_t //\nwuffs_base__u8__max(uint8_t x, uint8_t y) {\n return x > y ? x : y;\n}\n\nstatic inline uint16_t //\nwuffs_base__u16__min(uint16_t x, uint16_t y) {\n return x < y ? x : y;\n}\n\nstatic inline uint16_t //\nwuffs_base__u16__max(uint16_t x, uint16_t y) {\n return x > y ? x : y;\n}\n\nstatic inline uint32_t //\nwuffs_base__u32__min(uint32_t x, uint32_t y) {\n return x < y ? x : y;\n}\n" +
"\nstatic inline uint32_t //\nwuffs_base__u32__max(uint32_t x, uint32_t y) {\n return x > y ? x : y;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__min(uint64_t x, uint64_t y) {\n return x < y ? x : y;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__max(uint64_t x, uint64_t y) {\n return x > y ? x : y;\n}\n\n" +
"" +
"// --------\n\n// Saturating arithmetic (sat_add, sat_sub) branchless bit-twiddling algorithms\n// are per https://locklessinc.com/articles/sat_arithmetic/\n//\n// It is important that the underlying types are unsigned integers, as signed\n// integer arithmetic overflow is undefined behavior in C.\n\nstatic inline uint8_t //\nwuffs_base__u8__sat_add(uint8_t x, uint8_t y) {\n uint8_t res = (uint8_t)(x + y);\n res |= (uint8_t)(-(res < x));\n return res;\n}\n\nstatic inline uint8_t //\nwuffs_base__u8__sat_sub(uint8_t x, uint8_t y) {\n uint8_t res = (uint8_t)(x - y);\n res &= (uint8_t)(-(res <= x));\n return res;\n}\n\nstatic inline uint16_t //\nwuffs_base__u16__sat_add(uint16_t x, uint16_t y) {\n uint16_t res = (uint16_t)(x + y);\n res |= (uint16_t)(-(res < x));\n return res;\n}\n\nstatic inline uint16_t //\nwuffs_base__u16__sat_sub(uint16_t x, uint16_t y) {\n uint16_t res = (uint16_t)(x - y);\n res &= (uint16_t)(-(res <= x));\n return res;\n}\n\nstatic inline uint32_t //\nwuffs_base__u32__sat_add(uint32_t x, uint32_t y) {\n uint32" +
"_t res = (uint32_t)(x + y);\n res |= (uint32_t)(-(res < x));\n return res;\n}\n\nstatic inline uint32_t //\nwuffs_base__u32__sat_sub(uint32_t x, uint32_t y) {\n uint32_t res = (uint32_t)(x - y);\n res &= (uint32_t)(-(res <= x));\n return res;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__sat_add(uint64_t x, uint64_t y) {\n uint64_t res = (uint64_t)(x + y);\n res |= (uint64_t)(-(res < x));\n return res;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__sat_sub(uint64_t x, uint64_t y) {\n uint64_t res = (uint64_t)(x - y);\n res &= (uint64_t)(-(res <= x));\n return res;\n}\n\n" +
"_t res = (uint32_t)(x + y);\n res |= (uint32_t)(-(res < x));\n return res;\n}\n\nstatic inline uint32_t //\nwuffs_base__u32__sat_sub(uint32_t x, uint32_t y) {\n uint32_t res = (uint32_t)(x - y);\n res &= (uint32_t)(-(res <= x));\n return res;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__sat_add(uint64_t x, uint64_t y) {\n uint64_t res = (uint64_t)(x + y);\n res |= (uint64_t)(-(res < x));\n return res;\n}\n\nstatic inline uint64_t //\nwuffs_base__u64__sat_sub(uint64_t x, uint64_t y) {\n uint64_t res = (uint64_t)(x - y);\n res &= (uint64_t)(-(res <= x));\n return res;\n}\n\n " +
"" +
"// --------\n\n#define wuffs_base__load_u8be__no_bounds_check \\\n wuffs_base__load_u8__no_bounds_check\n#define wuffs_base__load_u8le__no_bounds_check \\\n wuffs_base__load_u8__no_bounds_check\n\nstatic inline uint8_t //\nwuffs_base__load_u8__no_bounds_check(uint8_t* p) {\n return p[0];\n}\n\nstatic inline uint16_t //\nwuffs_base__load_u16be__no_bounds_check(uint8_t* p) {\n return (uint16_t)(((uint16_t)(p[0]) << 8) | ((uint16_t)(p[1]) << 0));\n}\n\nstatic inline uint16_t //\nwuffs_base__load_u16le__no_bounds_check(uint8_t* p) {\n return (uint16_t)(((uint16_t)(p[0]) << 0) | ((uint16_t)(p[1]) << 8));\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u24be__no_bounds_check(uint8_t* p) {\n return ((uint32_t)(p[0]) << 16) | ((uint32_t)(p[1]) << 8) |\n ((uint32_t)(p[2]) << 0);\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u24le__no_bounds_check(uint8_t* p) {\n return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]) << 8) |\n ((uint32_t)(p[2]) << 16);\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u32be__no_bounds_check(" +
"uint8_t* p) {\n return ((uint32_t)(p[0]) << 24) | ((uint32_t)(p[1]) << 16) |\n ((uint32_t)(p[2]) << 8) | ((uint32_t)(p[3]) << 0);\n}\n\nstatic inline uint32_t //\nwuffs_base__load_u32le__no_bounds_check(uint8_t* p) {\n return ((uint32_t)(p[0]) << 0) | ((uint32_t)(p[1]) << 8) |\n ((uint32_t)(p[2]) << 16) | ((uint32_t)(p[3]) << 24);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u40be__no_bounds_check(uint8_t* p) {\n return ((uint64_t)(p[0]) << 32) | ((uint64_t)(p[1]) << 24) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 8) |\n ((uint64_t)(p[4]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u40le__no_bounds_check(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u48be__no_bounds_check(uint8_t* p) {\n return ((uint64_t)(p[0]) << 40) | ((uint64_t)(p[1]) << 32) |\n ((uint64_t)(p[2]) << 24) | ((uint64_t)" +
"(p[3]) << 16) |\n ((uint64_t)(p[4]) << 8) | ((uint64_t)(p[5]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u48le__no_bounds_check(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u56be__no_bounds_check(uint8_t* p) {\n return ((uint64_t)(p[0]) << 48) | ((uint64_t)(p[1]) << 40) |\n ((uint64_t)(p[2]) << 32) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 16) | ((uint64_t)(p[5]) << 8) |\n ((uint64_t)(p[6]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u56le__no_bounds_check(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |\n ((uint64_t)(p[6]) << 48);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u64be__no_bounds_ch" +
"eck(uint8_t* p) {\n return ((uint64_t)(p[0]) << 56) | ((uint64_t)(p[1]) << 48) |\n ((uint64_t)(p[2]) << 40) | ((uint64_t)(p[3]) << 32) |\n ((uint64_t)(p[4]) << 24) | ((uint64_t)(p[5]) << 16) |\n ((uint64_t)(p[6]) << 8) | ((uint64_t)(p[7]) << 0);\n}\n\nstatic inline uint64_t //\nwuffs_base__load_u64le__no_bounds_check(uint8_t* p) {\n return ((uint64_t)(p[0]) << 0) | ((uint64_t)(p[1]) << 8) |\n ((uint64_t)(p[2]) << 16) | ((uint64_t)(p[3]) << 24) |\n ((uint64_t)(p[4]) << 32) | ((uint64_t)(p[5]) << 40) |\n ((uint64_t)(p[6]) << 48) | ((uint64_t)(p[7]) << 56);\n}\n\n " +
"" +
"// --------\n\n#define wuffs_base__store_u8be__no_bounds_check \\\n wuffs_base__store_u8__no_bounds_check\n#define wuffs_base__store_u8le__no_bounds_check \\\n wuffs_base__store_u8__no_bounds_check\n\nstatic inline void //\nwuffs_base__store_u8__no_bounds_check(uint8_t* p, uint8_t x) {\n p[0] = x;\n}\n\nstatic inline void //\nwuffs_base__store_u16be__no_bounds_check(uint8_t* p, uint16_t x) {\n p[0] = (uint8_t)(x >> 8);\n p[1] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u16le__no_bounds_check(uint8_t* p, uint16_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n}\n\nstatic inline void //\nwuffs_base__store_u24be__no_bounds_check(uint8_t* p, uint32_t x) {\n p[0] = (uint8_t)(x >> 16);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u24le__no_bounds_check(uint8_t* p, uint32_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n}\n\nstatic inline void //\nwuffs_base__store_u32be__no_bounds_check(uint8_t* p," +
" uint32_t x) {\n p[0] = (uint8_t)(x >> 24);\n p[1] = (uint8_t)(x >> 16);\n p[2] = (uint8_t)(x >> 8);\n p[3] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u32le__no_bounds_check(uint8_t* p, uint32_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n}\n\nstatic inline void //\nwuffs_base__store_u40be__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 32);\n p[1] = (uint8_t)(x >> 24);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 8);\n p[4] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u40le__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32);\n}\n\nstatic inline void //\nwuffs_base__store_u48be__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 40);\n p[1] = (uint8_t)(x >> 32);\n p[2] = (uint8_t)(x >> 24);\n p[3] = (uint8_t)(x >> 16);\n p[4] " +
"= (uint8_t)(x >> 8);\n p[5] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u48le__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32);\n p[5] = (uint8_t)(x >> 40);\n}\n\nstatic inline void //\nwuffs_base__store_u56be__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 48);\n p[1] = (uint8_t)(x >> 40);\n p[2] = (uint8_t)(x >> 32);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 16);\n p[5] = (uint8_t)(x >> 8);\n p[6] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u56le__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32);\n p[5] = (uint8_t)(x >> 40);\n p[6] = (uint8_t)(x >> 48);\n}\n\nstatic inline void //\nwuffs_base__store_u64be__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 56);\n" +
" p[1] = (uint8_t)(x >> 48);\n p[2] = (uint8_t)(x >> 40);\n p[3] = (uint8_t)(x >> 32);\n p[4] = (uint8_t)(x >> 24);\n p[5] = (uint8_t)(x >> 16);\n p[6] = (uint8_t)(x >> 8);\n p[7] = (uint8_t)(x >> 0);\n}\n\nstatic inline void //\nwuffs_base__store_u64le__no_bounds_check(uint8_t* p, uint64_t x) {\n p[0] = (uint8_t)(x >> 0);\n p[1] = (uint8_t)(x >> 8);\n p[2] = (uint8_t)(x >> 16);\n p[3] = (uint8_t)(x >> 24);\n p[4] = (uint8_t)(x >> 32);\n p[5] = (uint8_t)(x >> 40);\n p[6] = (uint8_t)(x >> 48);\n p[7] = (uint8_t)(x >> 56);\n}\n\n" +
"" +
"// ---------------- Slices and Tables\n\n// WUFFS_BASE__SLICE is a 1-dimensional buffer.\n//\n// len measures a number of elements, not necessarily a size in bytes.\n//\n// A value with all fields NULL or zero is a valid, empty slice.\n#define WUFFS_BASE__SLICE(T) \\\n struct { \\\n T* ptr; \\\n size_t len; \\\n }\n\n// WUFFS_BASE__TABLE is a 2-dimensional buffer.\n//\n// width height, and stride measure a number of elements, not necessarily a\n// size in bytes.\n//\n// A value with all fields NULL or zero is a valid, empty table.\n#define WUFFS_BASE__TABLE(T) \\\n struct { \\\n T* ptr; \\\n size_t width; \\\n size_t height; \\\n size_t stride; \\\n }\n\ntypedef WUFFS_BASE__SLICE(uint8_t) wuffs_base__slice_u8;\ntypedef WUFFS_BASE__SLICE(uint16_t) wuffs_base__slice_u16;\ntypedef WUFFS_BASE__SLICE(uint32_t) wuffs_base__slice_u32;\ntypedef WUFFS_BASE__SLICE(uint64_t) wuffs_base__slice_u64;\n\ntypedef WUFFS_BASE__TABLE(u" +
"int8_t) wuffs_base__table_u8;\ntypedef WUFFS_BASE__TABLE(uint16_t) wuffs_base__table_u16;\ntypedef WUFFS_BASE__TABLE(uint32_t) wuffs_base__table_u32;\ntypedef WUFFS_BASE__TABLE(uint64_t) wuffs_base__table_u64;\n\nstatic inline wuffs_base__slice_u8 //\nwuffs_base__make_slice_u8(uint8_t* ptr, size_t len) {\n wuffs_base__slice_u8 ret;\n ret.ptr = ptr;\n ret.len = len;\n return ret;\n}\n\nstatic inline wuffs_base__slice_u16 //\nwuffs_base__make_slice_u16(uint16_t* ptr, size_t len) {\n wuffs_base__slice_u16 ret;\n ret.ptr = ptr;\n ret.len = len;\n return ret;\n}\n\nstatic inline wuffs_base__slice_u32 //\nwuffs_base__make_slice_u32(uint32_t* ptr, size_t len) {\n wuffs_base__slice_u32 ret;\n ret.ptr = ptr;\n ret.len = len;\n return ret;\n}\n\nstatic inline wuffs_base__slice_u64 //\nwuffs_base__make_slice_u64(uint64_t* ptr, size_t len) {\n wuffs_base__slice_u64 ret;\n ret.ptr = ptr;\n ret.len = len;\n return ret;\n}\n\nstatic inline wuffs_base__slice_u8 //\nwuffs_base__empty_slice_u8() {\n wuffs_base__slice_u8 ret;\n ret.ptr = NULL;\n" +
@@ -216,8 +218,8 @@ const baseIOPrivateH = "" +
" that:\n// - distance > 0\n// - distance <= (*ptr_iop_w - io1_w)\n// - length <= (io2_w - *ptr_iop_w)\nstatic inline uint32_t //\nwuffs_base__io_writer__copy_n32_from_history_fast(uint8_t** ptr_iop_w,\n uint8_t* io1_w,\n uint8_t* io2_w,\n uint32_t length,\n uint32_t distance) {\n uint8_t* p = *ptr_iop_w;\n uint8_t* q = p - distance;\n uint32_t n = length;\n for (; n >= 3; n -= 3) {\n *p++ = *q++;\n *p++ = *q++;\n *p++ = *q++;\n }\n for (; n; n--) {\n *p++ = *q++;\n }\n *ptr_iop_w = p;\n return length;\n}\n\nstatic inline uint32_t //\nwuffs_base__io_writer__copy_n32_from_reader(uint8_t** ptr_iop_w,\n uint8_t* io2_w,\n uint32_t length,\n uint8_t** ptr_iop_r,\n " +
" uint8_t* io2_r) {\n uint8_t* iop_w = *ptr_iop_w;\n size_t n = length;\n if (n > ((size_t)(io2_w - iop_w))) {\n n = (size_t)(io2_w - iop_w);\n }\n uint8_t* iop_r = *ptr_iop_r;\n if (n > ((size_t)(io2_r - iop_r))) {\n n = (size_t)(io2_r - iop_r);\n }\n if (n > 0) {\n memmove(iop_w, iop_r, n);\n *ptr_iop_w += n;\n *ptr_iop_r += n;\n }\n return (uint32_t)(n);\n}\n\nstatic inline uint64_t //\nwuffs_base__io_writer__copy_from_slice(uint8_t** ptr_iop_w,\n uint8_t* io2_w,\n wuffs_base__slice_u8 src) {\n uint8_t* iop_w = *ptr_iop_w;\n size_t n = src.len;\n if (n > ((size_t)(io2_w - iop_w))) {\n n = (size_t)(io2_w - iop_w);\n }\n if (n > 0) {\n memmove(iop_w, src.ptr, n);\n *ptr_iop_w += n;\n }\n return (uint64_t)(n);\n}\n\nstatic inline uint32_t //\nwuffs_base__io_writer__copy_n32_from_slice(uint8_t** ptr_iop_w,\n uint8_t* io2_w,\n uint32_t" +
" length,\n wuffs_base__slice_u8 src) {\n uint8_t* iop_w = *ptr_iop_w;\n size_t n = src.len;\n if (n > length) {\n n = length;\n }\n if (n > ((size_t)(io2_w - iop_w))) {\n n = (size_t)(io2_w - iop_w);\n }\n if (n > 0) {\n memmove(iop_w, src.ptr, n);\n *ptr_iop_w += n;\n }\n return (uint32_t)(n);\n}\n\n// wuffs_base__io_reader__match7 returns whether the io_reader's upcoming bytes\n// start with the given prefix (up to 7 bytes long). It is peek-like, not\n// read-like, in that there are no side-effects.\n//\n// The low 3 bits of a hold the prefix length, n.\n//\n// The high 56 bits of a hold the prefix itself, in little-endian order. The\n// first prefix byte is in bits 8..=15, the second prefix byte is in bits\n// 16..=23, etc. The high (8 * (7 - n)) bits are ignored.\n//\n// There are three possible return values:\n// - 0 means success.\n// - 1 means inconclusive, equivalent to \"$short read\".\n// - 2 means failure.\nstatic inline uint32_t //\nwuffs_base__io_reader__match7(ui" +
"nt8_t* iop_r,\n uint8_t* io2_r,\n wuffs_base__io_buffer* r,\n uint64_t a) {\n uint32_t n = a & 7;\n a >>= 8;\n if ((io2_r - iop_r) >= 8) {\n uint64_t x = wuffs_base__load_u64le(iop_r);\n uint32_t shift = 8 * (8 - n);\n return ((a << shift) == (x << shift)) ? 0 : 2;\n }\n for (; n > 0; n--) {\n if (iop_r >= io2_r) {\n return (r && r->meta.closed) ? 2 : 1;\n } else if (*iop_r != ((uint8_t)(a))) {\n return 2;\n }\n iop_r++;\n a >>= 8;\n }\n return 0;\n}\n\nstatic inline wuffs_base__io_buffer* //\nwuffs_base__io_reader__set(wuffs_base__io_buffer* b,\n uint8_t** ptr_iop_r,\n uint8_t** ptr_io0_r,\n uint8_t** ptr_io1_r,\n uint8_t** ptr_io2_r,\n wuffs_base__slice_u8 data) {\n b->data = data;\n b->meta.wi = data.len;\n b->meta.ri = 0;\n b->meta.pos = 0;\n b->meta.closed = false;\n\n *ptr_iop_r" +
" = data.ptr;\n *ptr_io0_r = data.ptr;\n *ptr_io1_r = data.ptr;\n *ptr_io2_r = data.ptr + data.len;\n\n return b;\n}\n\nstatic inline wuffs_base__slice_u8 //\nwuffs_base__io_reader__take(uint8_t** ptr_iop_r, uint8_t* io2_r, uint64_t n) {\n if (n <= ((size_t)(io2_r - *ptr_iop_r))) {\n uint8_t* p = *ptr_iop_r;\n *ptr_iop_r += n;\n return wuffs_base__make_slice_u8(p, n);\n }\n return wuffs_base__make_slice_u8(NULL, 0);\n}\n\nstatic inline wuffs_base__io_buffer* //\nwuffs_base__io_writer__set(wuffs_base__io_buffer* b,\n uint8_t** ptr_iop_w,\n uint8_t** ptr_io0_w,\n uint8_t** ptr_io1_w,\n uint8_t** ptr_io2_w,\n wuffs_base__slice_u8 data) {\n b->data = data;\n b->meta.wi = 0;\n b->meta.ri = 0;\n b->meta.pos = 0;\n b->meta.closed = false;\n\n *ptr_iop_w = data.ptr;\n *ptr_io0_w = data.ptr;\n *ptr_io1_w = data.ptr;\n *ptr_io2_w = data.ptr + data.len;\n\n return b;\n}\n\n" +
"nt8_t* iop_r,\n uint8_t* io2_r,\n wuffs_base__io_buffer* r,\n uint64_t a) {\n uint32_t n = a & 7;\n a >>= 8;\n if ((io2_r - iop_r) >= 8) {\n uint64_t x = wuffs_base__load_u64le__no_bounds_check(iop_r);\n uint32_t shift = 8 * (8 - n);\n return ((a << shift) == (x << shift)) ? 0 : 2;\n }\n for (; n > 0; n--) {\n if (iop_r >= io2_r) {\n return (r && r->meta.closed) ? 2 : 1;\n } else if (*iop_r != ((uint8_t)(a))) {\n return 2;\n }\n iop_r++;\n a >>= 8;\n }\n return 0;\n}\n\nstatic inline wuffs_base__io_buffer* //\nwuffs_base__io_reader__set(wuffs_base__io_buffer* b,\n uint8_t** ptr_iop_r,\n uint8_t** ptr_io0_r,\n uint8_t** ptr_io1_r,\n uint8_t** ptr_io2_r,\n wuffs_base__slice_u8 data) {\n b->data = data;\n b->meta.wi = data.len;\n b->meta.ri = 0;\n b->meta.pos = 0;\n b->meta.closed = fal" +
"se;\n\n *ptr_iop_r = data.ptr;\n *ptr_io0_r = data.ptr;\n *ptr_io1_r = data.ptr;\n *ptr_io2_r = data.ptr + data.len;\n\n return b;\n}\n\nstatic inline wuffs_base__slice_u8 //\nwuffs_base__io_reader__take(uint8_t** ptr_iop_r, uint8_t* io2_r, uint64_t n) {\n if (n <= ((size_t)(io2_r - *ptr_iop_r))) {\n uint8_t* p = *ptr_iop_r;\n *ptr_iop_r += n;\n return wuffs_base__make_slice_u8(p, n);\n }\n return wuffs_base__make_slice_u8(NULL, 0);\n}\n\nstatic inline wuffs_base__io_buffer* //\nwuffs_base__io_writer__set(wuffs_base__io_buffer* b,\n uint8_t** ptr_iop_w,\n uint8_t** ptr_io0_w,\n uint8_t** ptr_io1_w,\n uint8_t** ptr_io2_w,\n wuffs_base__slice_u8 data) {\n b->data = data;\n b->meta.wi = 0;\n b->meta.ri = 0;\n b->meta.pos = 0;\n b->meta.closed = false;\n\n *ptr_iop_w = data.ptr;\n *ptr_io0_w = data.ptr;\n *ptr_io1_w = data.ptr;\n *ptr_io2_w = data.ptr + data.len;\n\n return b;\n}\n\n " +
"" +
"// ---------------- I/O (Utility)\n\n#define wuffs_base__utility__empty_io_reader wuffs_base__empty_io_reader\n#define wuffs_base__utility__empty_io_writer wuffs_base__empty_io_writer\n" +
""