From 565c1562c667dbbcf46aaaa73122ceeff0216323 Mon Sep 17 00:00:00 2001 From: Tenari Date: Sat, 31 Jan 2026 09:59:03 -0600 Subject: [PATCH] empty client-server game --- .gitignore | 2 + make.sh | 47 + src/assets/.keep | 0 src/base/all.h | 599 +++ src/base/entry.c | 45 + src/base/impl.c | 12 + src/base/linux_os.c | 139 + src/base/mac_os.c | 136 + src/base/math.c | 68 + src/base/memory.c | 94 + src/base/os.c | 12 + src/base/pthread_barrier.c | 102 + src/base/pthread_barrier.h | 49 + src/base/serialize.c | 66 + src/base/string.c | 216 + src/base/tctx.c | 83 + src/base/thread.c | 36 + src/base/thread_context.h | 114 + src/base/unix_os.c | 147 + src/base/win32_os.c | 326 ++ src/client.c | 743 ++++ src/lib/network.c | 198 + src/lib/stb_image.h | 7988 ++++++++++++++++++++++++++++++++++++ src/lib/thread.c | 109 + src/lib/thread.h | 29 + src/lib/tui.c | 579 +++ src/render.c | 104 + src/server.c | 785 ++++ src/shared.h | 77 + src/string_chunk.c | 162 + src/string_chunk.h | 33 + win_make.bat | 4 + 32 files changed, 13104 insertions(+) create mode 100644 .gitignore create mode 100755 make.sh create mode 100644 src/assets/.keep create mode 100644 src/base/all.h create mode 100644 src/base/entry.c create mode 100644 src/base/impl.c create mode 100644 src/base/linux_os.c create mode 100644 src/base/mac_os.c create mode 100644 src/base/math.c create mode 100644 src/base/memory.c create mode 100644 src/base/os.c create mode 100644 src/base/pthread_barrier.c create mode 100644 src/base/pthread_barrier.h create mode 100644 src/base/serialize.c create mode 100644 src/base/string.c create mode 100644 src/base/tctx.c create mode 100644 src/base/thread.c create mode 100644 src/base/thread_context.h create mode 100644 src/base/unix_os.c create mode 100644 src/base/win32_os.c create mode 100644 src/client.c create mode 100644 src/lib/network.c create mode 100644 src/lib/stb_image.h create mode 100644 src/lib/thread.c create mode 100644 src/lib/thread.h create mode 100644 src/lib/tui.c create mode 100644 src/render.c create mode 100644 src/server.c create mode 100644 src/shared.h create mode 100644 src/string_chunk.c create mode 100644 src/string_chunk.h create mode 100644 win_make.bat diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000..85f404f --- /dev/null +++ b/.gitignore @@ -0,0 +1,2 @@ +*.swp +build/* diff --git a/make.sh b/make.sh new file mode 100755 index 0000000..576f075 --- /dev/null +++ b/make.sh @@ -0,0 +1,47 @@ +#!/bin/sh + +if [ "$1" = "server" ]; then + echo "building server" + rm ./build/server + rm -rf ./build/server.dSYM + if [[ $(uname) == "Darwin" ]]; then + echo " for macOS" + gcc -std=c99 -g -o build/server src/server.c -lpthread + else + echo " for Linux" + gcc -std=c99 -D_POSIX_C_SOURCE=200809L -g -o build/server src/server.c -lpthread + fi + if [ "$2" = "run" ]; then + ./build/server + elif [ "$2" = "debug-run" ]; then + ../lldbg/bin/lldbgui ./build/server + fi +elif [ "$1" = "editor" ]; then + echo "building editor" + rm ./build/editor + rm -rf ./build/editor.dSYM + gcc -std=c99 -g -o build/editor src/room_editor.c + if [ "$2" = "run" ]; then + ./build/editor $3 + fi +else + #echo "building assets" + #for item in "asset1" "asset2" "asset3"; do + # rm "src/assets/$item.h" + # xxd -i "assets/$item.txt" > "src/assets/$item.h" + #done + + echo "building client" + rm ./build/client + rm -rf ./build/client.dSYM + gcc -std=c99 -g -o build/client src/client.c -lpthread + if [ "$2" = "run" ]; then + if [ "$3" ]; then + ./build/client $3 + else + ./build/client + fi + elif [ "$2" = "debug-run" ]; then + ../lldbg/bin/lldbgui ./build/client + fi +fi diff --git a/src/assets/.keep b/src/assets/.keep new file mode 100644 index 0000000..e69de29 diff --git a/src/base/all.h b/src/base/all.h new file mode 100644 index 0000000..52a86a0 --- /dev/null +++ b/src/base/all.h @@ -0,0 +1,599 @@ +// good base layer overview https://www.youtube.com/watch?v=bUOOaXf9qIM +#ifndef BASE_ALL_H +#define BASE_ALL_H + +///// Context Cracking +// Development Settings +#if !defined(ENABLE_ASSERT) +# define ENABLE_ASSERT 1 +#endif +#if !defined(ENABLE_SANITIZER) +# define ENABLE_SANITIZER 0 +#endif +#if !defined(ENABLE_MANUAL_PROFILE) +# define ENABLE_MANUAL_PROFILE 0 +#endif +#if !defined(ENABLE_AUTO_PROFILE) +# define ENABLE_AUTO_PROFILE 0 +#endif + +#if defined(ENABLE_ANY_PROFILE) +# error user should not configure ENABLE_ANY_PROFILE +#endif + +#if ENABLE_MANUAL_PROFILE || ENABLE_AUTO_PROFILE +# define ENABLE_ANY_PROFILE 1 +#else +# define ENABLE_ANY_PROFILE 0 +#endif + +// Untangle Compiler, OS & Architecture +#if defined(__clang__) +# define COMPILER_CLANG 1 + +# if defined(_WIN32) +# define OS_WINDOWS 1 +# elif defined(__gnu_linux__) +# define OS_LINUX 1 +# elif defined(__APPLE__) && defined(__MACH__) +# define OS_MAC 1 +# else +# error missing OS detection +# endif + +# if defined(__amd64__) +# define ARCH_X64 1 +// TODO verify this works on clang +# elif defined(__i386__) +# define ARCH_X86 1 +// TODO verify this works on clang +# elif defined(__arm__) +# define ARCH_ARM 1 +// TODO verify this works on clang +# elif defined(__aarch64__) +# define ARCH_ARM64 1 +# else +# error missing ARCH detection +# endif + +#elif defined(_MSC_VER) +# define COMPILER_CL 1 + +# if defined(_WIN32) +# define OS_WINDOWS 1 +# else +# error missing OS detection +# endif + +# if defined(_M_AMD64) +# define ARCH_X64 1 +# elif defined(_M_I86) +# define ARCH_X86 1 +# elif defined(_M_ARM) +# define ARCH_ARM 1 +// TODO ARM64? +# else +# error missing ARCH detection +# endif + +#elif defined(__GNUC__) +# define COMPILER_GCC 1 + +# if defined(_WIN32) +# define OS_WINDOWS 1 +# elif defined(__gnu_linux__) +# define OS_LINUX 1 +# elif defined(__APPLE__) && defined(__MACH__) +# define OS_MAC 1 +# else +# error missing OS detection +# endif + +# if defined(__amd64__) +# define ARCH_X64 1 +# elif defined(__i386__) +# define ARCH_X86 1 +# elif defined(__arm__) +# define ARCH_ARM 1 +# elif defined(__aarch64__) +# define ARCH_ARM64 1 +# else +# error missing ARCH detection +# endif + +#else +# error no context cracking for this compiler +#endif + +#if !defined(COMPILER_CL) +# define COMPILER_CL 0 +#endif +#if !defined(COMPILER_CLANG) +# define COMPILER_CLANG 0 +#endif +#if !defined(COMPILER_GCC) +# define COMPILER_GCC 0 +#endif +#if !defined(OS_WINDOWS) +# define OS_WINDOWS 0 +#endif +#if !defined(OS_LINUX) +# define OS_LINUX 0 +#endif +#if !defined(OS_MAC) +# define OS_MAC 0 +#endif +#if !defined(ARCH_X64) +# define ARCH_X64 0 +#endif +#if !defined(ARCH_X86) +# define ARCH_X86 0 +#endif +#if !defined(ARCH_ARM) +# define ARCH_ARM 0 +#endif +#if !defined(ARCH_ARM64) +# define ARCH_ARM64 0 +#endif + +// Language +#if defined(__cplusplus) +# define LANG_CXX 1 +#else +# define LANG_C 1 +#endif + +#if !defined(LANG_CXX) +# define LANG_CXX 0 +#endif +#if !defined(LANG_C) +# define LANG_C 0 +#endif + +// Profiler +#if !defined(PROFILER_SPALL) +# define PROFILER_SPALL 0 +#endif + +// Determine Intrinsics Mode +#if OS_WINDOWS +# if COMPILER_CL || COMPILER_CLANG +# define INTRINSICS_MICROSOFT 1 +# endif +#endif + +#if !defined(INTRINSICS_MICROSOFT) +# define INTRINSICS_MICROSOFT 0 +#endif + +// Setup Pointer Size Macro +#if ARCH_X64 || ARCH_ARM64 +# define ARCH_ADDRSIZE 64 +#else +# define ARCH_ADDRSIZE 32 +#endif + +///// MACROS +#define global static +#define internal static +#define fn static + +#define arrayLen(a) (sizeof(a)/sizeof(*(a))) + +#define intFromPtr(p) (U64)((U8*)p - (U8*)0) +#define ptrFromInt(n) (void*)((U8*)0 + (n)) + +#define stmnt(S) do{ S }while(0) + +#if !defined(assertBreak) +# define assertBreak() (*(volatile int*)0 = 0) +#endif + +#if ENABLE_ASSERT +# define assert(c) stmnt( if (!(c)){ assertBreak(); } ) +#else +# define assert(c) +#endif + +#define ToBool(x) ((x) != 0) + +#define KB(x) ((x) << 10) +#define MB(x) ((x) << 20) +#define GB(x) ((x) << 30) +#define TB(x) ((u64)(x) << 40llu) + +#include +#define MemoryCopy(d,s,z) memmove((d), (s), (z)) +#define MemoryCopyStruct(d,s) MemoryCopy((d),(s), Min(sizeof(*(d)) , sizeof(*(s)))) +#define MemoryZero(d,z) memset((d), 0, (z)) +#define MemoryZeroStruct(d,s) MemoryZero((d),sizeof(s)) +#define Min(a,b) (((a)<(b))?(a):(b)) +#define Max(a,b) (((a)>(b))?(a):(b)) +#define Abs(x) (((x)<0)?((x)*-1):(x)) + +#define SetFlag(flags, bit) ((flags) |= (1ULL << (bit))) +#define ClearFlag(flags, bit) ((flags) &= ~(1ULL << (bit))) +#define ToggleFlag(flags, bit) ((flags) ^= (1ULL << (bit))) +#define CheckFlag(flags, bit) (((flags) >> (bit)) & 1ULL) + +#define QueuePush(f,l,n) (((f)==NULL) ? ((f)=(l)=(n)) : ((l)->next=(n),(l)=(n),(n)->next = NULL)) + +#define DEFAULT_ALIGNMENT sizeof(void*) +#define isPowerOfTwo(x) ((x & (x-1)) == 0) + +#define XYToPos(x, y, w) ((u32)(((u32)(x)) + (((u32)(y)) * (w)))) + +#if COMPILER_MSVC +# define thread_static __declspec(thread) +#elif COMPILER_CLANG || COMPILER_GCC +# define thread_static __thread +#else +# error thread_static not defined for this compiler. +#endif + +// only valid if there's an in-scope variable bool `debug_mode` +#define dbg(fmt, ...) osDebugPrint(debug_mode, fmt, ##__VA_ARGS__) + +///// TYPES +// integer types +typedef unsigned char u8; +typedef signed char i8; +typedef unsigned short u16; +typedef short i16; +typedef unsigned int u32; +typedef int i32; +typedef unsigned long long u64; +typedef signed long long i64; +typedef char * usize; +typedef char * ptr; +typedef const char* str; + +// Floating point types +typedef float f32; +typedef double f64; + +//typedef long double f80; only returning 8 bytes on my mac for some reason +/* + printf("u8 %d\n", (int)sizeof(u8) * 8); + printf("i8 %d\n", (int)sizeof(i8) * 8); + printf("u16 %d\n", (int)sizeof(u16) * 8); + printf("i16 %d\n", (int)sizeof(i16) * 8); + printf("u32 %d\n", (int)sizeof(u32) * 8); + printf("i32 %d\n", (int)sizeof(i32) * 8); + printf("u64 %d\n", (int)sizeof(u64) * 8); + printf("i64 %d\n", (int)sizeof(i64) * 8); + printf("usize %d\n", (int)sizeof(usize) * 8); + + printf("f32 %d\n", (int)sizeof(f32) * 8); + printf("f64 %d\n", (int)sizeof(f64) * 8); + printf("f80 %d\n", (int)sizeof(f80) * 8); +*/ + +// Boolean types +typedef u8 b8; +typedef u32 b32; +#ifndef bool +# define bool b8 +#endif + +#define true 1 +#define false 0 + +// Structs +typedef struct Arena { + u8* memory; + u64 max; + u64 alloc_position; + u64 commit_position; + b8 static_size; +} Arena; + +typedef struct PtrArray { + u32 length; + u32 capacity; + ptr* items; +} PtrArray; + +typedef struct u8List { + u32 length; + u32 capacity; + u8* items; +} u8List; + +typedef struct String { + u32 length; + u32 capacity; + ptr bytes; +} String; + +typedef struct StringUTF16Const { + u16* string; + u64 size; +} StringUTF16Const; + +typedef enum Utf8Character { + Utf8CharacterAscii, + Utf8CharacterTwoByte, + Utf8CharacterThreeByte, + Utf8CharacterFourByte, + Utf8Character_Count, +} Utf8Character; + +typedef struct Box { + u32 x; + u32 y; + u32 height; + u32 width; +} Box; + +typedef struct Dim2 { + u16 height; + u16 width; +} Dim2; + +typedef struct Pos2 { + u16 x; + u16 y; +} Pos2; + +typedef union Range1u32 Range1u32; +union Range1u32 +{ + struct + { + u32 min; + u32 max; + }; + u32 v[2]; +}; + +typedef union Range1i32 Range1i32; +union Range1i32 +{ + struct + { + i32 min; + i32 max; + }; + i32 v[2]; +}; + +typedef union Range1u64 Range1u64; +union Range1u64 +{ + struct + { + u64 min; + u64 max; + }; + u64 v[2]; +}; + +typedef union Range1i64 Range1i64; +union Range1i64 +{ + struct + { + i64 min; + i64 max; + }; + i64 v[2]; +}; + +typedef union Range1f32 Range1f32; +union Range1f32 +{ + struct + { + f32 min; + f32 max; + }; + f32 v[2]; +}; + +#if OS_WINDOWS +# define WIN32_LEAN_AND_MEAN +# include +# include +# include +# include +# include +# include + typedef struct { + DWORD input_mode; + DWORD output_mode; + } TermIOs; +#else +# include +# include +# include +# include +# include +# include +# include +# include +# include + typedef struct termios TermIOs; +#endif + +typedef struct Barrier { + u64 a[1]; +} Barrier; + +typedef struct CondVar { + u64 a[1]; +} CondVar; + +typedef struct Thread { + pthread_t thread; +} Thread; + +typedef struct Mutex { + pthread_mutex_t mutex; +} Mutex; + +typedef struct Cond { + pthread_cond_t cond; +} Cond; + +#define M_SCRATCH_SIZE KB(16) +typedef struct ScratchFreeListNode ScratchFreeListNode; +struct ScratchFreeListNode { + ScratchFreeListNode* next; + u32 index; +}; + +typedef struct ScratchMem { + Arena arena; + u32 index; + u64 pos; +} ScratchMem; + +typedef struct LaneCtx { + u64 lane_idx; + u64 lane_count; + Barrier barrier; + u64 *broadcast_memory; +} LaneCtx; + +typedef struct ThreadContext { + Arena arena; // scratch + u32 max_created; + ScratchFreeListNode* free_list; + LaneCtx lane_ctx; +} ThreadContext; + +///// HARDCODED GLOBALS +global const u64 MAX_u64 = 0xffffffffffffffffull; +global const u32 MAX_u32 = 0xffffffff; +global const u16 MAX_u16 = 0xffff; +global const u8 MAX_u8 = 0xff; + +///// CUSTOM ENTRY POINT +/* TODO? +fn void mainThreadBaseEntryPoint(i32 argc, char **argv); +fn void asyncThreadEntryPoint(void *params); +fn void supplement_thread_base_entry_point(void (*entry_point)(void *params), void *params); +fn u64 update_tick_idx(void); +fn b32 update(void); +*/ + +///// MATH +fn Range1u64 range1u64Create(u64 min, u64 max); +fn Range1u64 mRangeFromNIdxMCount(u64 n_idx, u64 n_count, u64 m_count); +fn void u32Quicksort(u32 arr[], u32 low, u32 high); +fn void u32ReverseArray(u32 arr[], u32 size); + +///// MEMORY (Arenas) +#define ARENA_MAX GB(1) +// this was an evil bug to figure out +#if defined(OS_MAC) +# define ARENA_COMMIT_SIZE KB(16) +#else +# define ARENA_COMMIT_SIZE KB(8) +#endif + +fn void* arenaAlloc(Arena* arena, u64 size); +fn void* arenaAllocZero(Arena* arena, u64 size); +fn void arenaDealloc(Arena* arena, u64 size); +fn void arenaDeallocTo(Arena* arena, u64 pos); +fn void* arenaRaise(Arena* arena, void* ptr, u64 size); +fn void* arenaAllocArraySized(Arena* arena, u64 elem_size, u64 count); +#define arenaAllocArray(arena, elem_type, count) arenaAllocArraySized(arena, sizeof(elem_type), count) + +fn void arenaInit(Arena* arena); +fn void arenaInitSized(Arena* arena, u64 max); +fn void arenaClear(Arena* arena); +fn void arenaFree(Arena* arena); + +ScratchMem scratchGet(void); +void scratchReset(ScratchMem* scratch); +void scratchReturn(ScratchMem* scratch); + +///// STRINGS +#define ASCII_TAB (9) +#define ASCII_LINE_FEED (10) +#define ASCII_RETURN (13) +#define ASCII_ESCAPE (27) +#define ASCII_DEL (127) +#define ASCII_BACKSPACE (8) + +fn bool stringsEq(String* a, String* b); +fn bool cStringEqString(str a, String* b); +fn Utf8Character classifyUtf8Character(u8 c); +fn bool isUtf8Ascii(u8 c); +fn bool isUtf8TwoByte(u8 c); +fn bool isUtf8ThreeByte(u8 c); +fn bool isUtf8FourByte(u8 c); +fn u8 lowerAscii(u8 c); +fn u8 upperAscii(u8 c); +fn StringUTF16Const str16FromStr8(Arena* a, String string); +fn bool isAlphaUnderscoreSpace(u8 c); +fn bool isSimplePrintable(u8 c); + +///// OS-wrapped apis +void osInit(); +void* osThreadContextGet(); +void osThreadContextSet(void* ctx); + +fn Barrier osBarrierAlloc(u64 count); +fn void osBarrierRelease(Barrier barrier); +fn void osBarrierWait(Barrier barrier); + +// Memory +fn void* osMemoryReserve(u64 size); +fn void osMemoryCommit(void* memory, u64 size); +fn void osMemoryDecommit(void* memory, u64 size); +fn void osMemoryRelease(void* memory, u64 size); +fn u64 osTimeMicrosecondsNow(); +fn void osSleepMicroseconds(u32 t); + +fn bool osFileExists(String filename); +fn String osFileRead(Arena* arena, ptr filepath); +fn bool osFileCreate(String filename); +fn bool osFileCreateWrite(String filename, String data); +fn bool osFileWrite(String filename, String data); + +fn void osDebugPrint(bool debug_mode, const char* format, ...); + +TermIOs osStartTUI(bool blocking); +fn void osEndTUI(TermIOs old_terminal_attributes); +fn Dim2 osGetTerminalDimensions(); +void osBlitToTerminal(ptr writeable_output_ansi_string, i64 count); +void osReadConsoleInput(u8* buffer, u32 len); + +bool osInitNetwork(); +i32 osLanIPAddress(); + +bool osThreadJoin(Thread handle, u64 endt_us); + +///// Basic THREAD synchronization apis +Thread spawnThread(void * (*threadFn)(void *), void* thread_arg); +Mutex newMutex(); +Cond newCond(); +void lockMutex(Mutex* m); +void unlockMutex(Mutex* m); +void signalCond(Cond* cond); +void waitForCondSignal(Cond* cond, Mutex* mutex); + +///// Multi-Core by Default ThreadContext stuff +void tctxInit(ThreadContext* ctx); +void tctxFree(ThreadContext* ctx); +fn ThreadContext *tctxSelected(void); + +ScratchMem tctxScratchGet(ThreadContext* ctx); +void tctxScratchReset(ThreadContext* ctx, ScratchMem* scratch); +void tctxScratchReturn(ThreadContext* ctx, ScratchMem* scratch); + +fn LaneCtx tctxSetLaneCtx(LaneCtx lane_ctx); +fn void tctxLaneBarrierWait(void *broadcast_ptr, u64 broadcast_size, u64 broadcast_src_lane_idx); +#define LaneIdx() (tctxSelected()->lane_ctx.lane_idx) +#define LaneCount() (tctxSelected()->lane_ctx.lane_count) +#define LaneFromTaskIdx(idx) ((idx)%LaneCount()) +#define LaneCtx(ctx) tctxSetLaneCtx((ctx)) +#define LaneSync() tctxLaneBarrierWait(0, 0, 0) +#define LaneSyncu64(pointer, src_lane_idx) tctxLaneBarrierWait((pointer), sizeof(*(pointer)), (src_lane_idx)) +#define LaneRange(count) mRangeFromNIdxMCount(LaneIdx(), LaneCount(), (count)) + + +#endif// BASE_ALL_H diff --git a/src/base/entry.c b/src/base/entry.c new file mode 100644 index 0000000..c38c5ef --- /dev/null +++ b/src/base/entry.c @@ -0,0 +1,45 @@ +#include "all.h" + +fn void mainThreadBaseEntryPoint(i32 argc, char **argv) { + Thread* async_threads = 0; + u64 lane_broadcast_val = 0; + { + u32 num_main_threads = 1; + u32 num_async_threads = 12;//TODO: os_get_system_info()->logical_processor_count; + u32 num_main_threads_clamped = Min(num_async_threads, num_main_threads); + num_async_threads -= num_main_threads_clamped; + String num_async_threads_string = cmd_line_string(&cmdline, str8_lit("async_thread_count")); + if(num_async_threads_string.size != 0) + { + try_u64_from_str8_c_rules(num_async_threads_string, &num_async_threads); + } + num_async_threads = Max(1, num_async_threads); + Barrier barrier = barrier_alloc(num_async_threads); + LaneCtx *lane_ctxs = push_array(scratch.arena, LaneCtx, num_async_threads); + async_threads_count = num_async_threads; + async_threads = push_array(scratch.arena, Thread, async_threads_count); + for EachIndex(idx, num_async_threads) + { + lane_ctxs[idx].lane_idx = idx; + lane_ctxs[idx].lane_count = async_threads_count; + lane_ctxs[idx].barrier = barrier; + lane_ctxs[idx].broadcast_memory = &lane_broadcast_val; + async_threads[idx] = thread_launch(async_thread_entry_point, &lane_ctxs[idx]); + } + } + + //- rjf: call into entry point + entry_point(&cmdline); + + //- rjf: join async threads + ins_atomic_u32_inc_eval(&global_async_exit); + cond_var_broadcast(async_tick_start_cond_var); + for EachIndex(idx, async_threads_count) + { + thread_join(async_threads[idx], max_U64); + } +} + +fn void asyncThreadEntryPoint(void *params) { +} + diff --git a/src/base/impl.c b/src/base/impl.c new file mode 100644 index 0000000..a01a259 --- /dev/null +++ b/src/base/impl.c @@ -0,0 +1,12 @@ +#ifndef BASE_IMPL_C +#define BASE_IMPL_C + +#include "math.c" +#include "os.c" +#include "memory.c" +#include "serialize.c" +#include "string.c" +#include "tctx.c" +#include "thread.c" + +#endif // BASE_IMPL_C diff --git a/src/base/linux_os.c b/src/base/linux_os.c new file mode 100644 index 0000000..b3f6f4b --- /dev/null +++ b/src/base/linux_os.c @@ -0,0 +1,139 @@ +#include +#define _POSIX_C_SOURCE 200809L +#include +#include +#include +#include +#include +#include +#include "all.h" + +global pthread_barrier_t linux_thread_barrier; + +fn Barrier osBarrierAlloc(u64 count) { + pthread_barrier_init(&linux_thread_barrier, NULL, count); + Barrier result = {(u64)&linux_thread_barrier}; + return result; +} + +fn void osBarrierRelease(Barrier barrier) { + pthread_barrier_t* addr = (pthread_barrier_t*)barrier.a[0]; + pthread_barrier_destroy(addr); +} + +fn void osBarrierWait(Barrier barrier) { + pthread_barrier_t* addr = (pthread_barrier_t*)barrier.a[0]; + pthread_barrier_wait(addr); +} + +// Time +fn u64 osTimeMicrosecondsNow() { + struct timespec ts; + clock_gettime(CLOCK_MONOTONIC, &ts); + return ((u64)ts.tv_sec * 1000000) + ((u64)ts.tv_nsec / 1000000); +} + +#define MICROSECONDS_PER_SECOND 1000000 +#define NANOSECONDS_PER_MICROSECOND 1000 +fn void osSleepMicroseconds(u32 t) { + struct timespec ts = { t / MICROSECONDS_PER_SECOND, (t % MICROSECONDS_PER_SECOND)*NANOSECONDS_PER_MICROSECOND }; + nanosleep(&ts, NULL); +} + +// Files +fn bool osFileExists(String filename) { + bool result = access((str)filename.bytes, F_OK) == 0; + return result; +} + +fn String osFileRead(Arena* arena, ptr filepath) { + struct stat st; + stat(filepath, &st); + String result = { st.st_size, st.st_size, 0 }; + result.bytes = arenaAlloc(arena, st.st_size); + + size_t handle = open(filepath, O_RDWR, S_IRUSR | S_IRGRP | S_IROTH); + read(handle, result.bytes, st.st_size); + close(handle); + + return result; +} + +fn bool osFileCreate(String filename) { + /* + M_Scratch scratch = scratch_get(); + string nt = str_copy(&scratch.arena, filename); + bool result = true; + size_t handle = open((const char*) nt.str, O_RDWR | O_CREAT, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) { + result = false; + } + scratch_return(&scratch); + close(handle); + return true; + */ + bool result = true; + size_t handle = open((str)filename.bytes, O_RDWR | O_CREAT, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) { + result = false; + } + if (close(handle) == -1) { + result = false; + } + return result; +} + +fn bool osFileCreateWrite(String filename, String data) { + /* + M_Scratch scratch = scratch_get(); + string nt = str_copy(&scratch.arena, filename); + b32 result = true; + size_t handle = + open((const char*) nt.str, O_RDWR | O_CREAT | O_TRUNC, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) result = false; + write(handle, data.str, data.size); + close(handle); + scratch_return(&scratch); + return result; + */ + bool result = true; + size_t handle = open( + (str)filename.bytes, + O_RDWR | O_CREAT | O_TRUNC, + S_IRUSR | S_IRGRP | S_IROTH + ); + if (handle == -1) result = false; + write(handle, data.bytes, data.length); + close(handle); + return result; +} + +fn bool osFileWrite(String filename, String data) { + /* + M_Scratch scratch = scratch_get(); + string nt = str_copy(&scratch.arena, filename); + b32 result = true; + size_t handle = + open((const char*) nt.str, O_RDWR | O_TRUNC, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) result = false; + write(handle, data.str, data.size); + close(handle); + */ + bool result = true; + size_t handle = open((str) filename.bytes, O_RDWR | O_TRUNC, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) result = false; + write(handle, data.bytes, data.length); + close(handle); + return result; +} + + +// Misc +fn void osDebugPrint(bool debug_mode, const char * format, ... ) { + if (debug_mode) { + va_list args; + va_start(args, format); + vprintf(format, args); + va_end(args); + } +} diff --git a/src/base/mac_os.c b/src/base/mac_os.c new file mode 100644 index 0000000..ede0437 --- /dev/null +++ b/src/base/mac_os.c @@ -0,0 +1,136 @@ +#include +#include +#include +#include +#include +#include +#include +#include "all.h" +#include "pthread_barrier.h" + +global pthread_barrier_t macos_thread_barrier; + +fn Barrier osBarrierAlloc(u64 count) { + pthread_barrier_init(&macos_thread_barrier, NULL, count); + Barrier result = {(u64)&macos_thread_barrier}; + return result; +} + +fn void osBarrierRelease(Barrier barrier) { + pthread_barrier_t* addr = (pthread_barrier_t*)barrier.a[0]; + pthread_barrier_destroy(addr); +} + +fn void osBarrierWait(Barrier barrier) { + pthread_barrier_t* addr = (pthread_barrier_t*)barrier.a[0]; + pthread_barrier_wait(addr); +} + +// Time +fn u64 osTimeMicrosecondsNow() { + struct timespec ts; + clock_gettime(CLOCK_MONOTONIC_RAW, &ts); + return ((u64)ts.tv_sec * 1000000) + ((u64)ts.tv_nsec / 1000000); +} + +fn void osSleepMicroseconds(u32 t) { + usleep(t); +} + +// Files +fn bool osFileExists(String filename) { + bool result = access((str)filename.bytes, F_OK) == 0; + return result; +} + +fn String osFileRead(Arena* arena, ptr filepath) { + struct stat st; + stat(filepath, &st); + String result = { st.st_size, st.st_size, 0 }; + result.bytes = arenaAlloc(arena, st.st_size); + + size_t handle = open(filepath, O_RDWR, S_IRUSR | S_IRGRP | S_IROTH); + read(handle, result.bytes, st.st_size); + close(handle); + + return result; +} + +fn bool osFileCreate(String filename) { + /* + M_Scratch scratch = scratch_get(); + string nt = str_copy(&scratch.arena, filename); + bool result = true; + size_t handle = open((const char*) nt.str, O_RDWR | O_CREAT, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) { + result = false; + } + scratch_return(&scratch); + close(handle); + return true; + */ + bool result = true; + size_t handle = open((str)filename.bytes, O_RDWR | O_CREAT, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) { + result = false; + } + if (close(handle) == -1) { + result = false; + } + return result; +} + +fn bool osFileCreateWrite(String filename, String data) { + /* + M_Scratch scratch = scratch_get(); + string nt = str_copy(&scratch.arena, filename); + b32 result = true; + size_t handle = + open((const char*) nt.str, O_RDWR | O_CREAT | O_TRUNC, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) result = false; + write(handle, data.str, data.size); + close(handle); + scratch_return(&scratch); + return result; + */ + bool result = true; + size_t handle = open( + (str)filename.bytes, + O_RDWR | O_CREAT | O_TRUNC, + S_IRUSR | S_IRGRP | S_IROTH + ); + if (handle == -1) result = false; + write(handle, data.bytes, data.length); + close(handle); + return result; +} + +fn bool osFileWrite(String filename, String data) { + /* + M_Scratch scratch = scratch_get(); + string nt = str_copy(&scratch.arena, filename); + b32 result = true; + size_t handle = + open((const char*) nt.str, O_RDWR | O_TRUNC, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) result = false; + write(handle, data.str, data.size); + close(handle); + */ + bool result = true; + size_t handle = open((str) filename.bytes, O_RDWR | O_TRUNC, S_IRUSR | S_IRGRP | S_IROTH); + if (handle == -1) result = false; + write(handle, data.bytes, data.length); + close(handle); + return result; +} + + +// Misc +fn void osDebugPrint(bool debug_mode, const char * format, ... ) { + if (debug_mode) { + va_list args; + va_start(args, format); + vprintf(format, args); + va_end(args); + } +} diff --git a/src/base/math.c b/src/base/math.c new file mode 100644 index 0000000..1035af0 --- /dev/null +++ b/src/base/math.c @@ -0,0 +1,68 @@ +#include "all.h" + +fn Range1u64 range1u64Create(u64 min, u64 max) { + Range1u64 result = { + .min = min, + .max = max + }; + if (result.min > result.max) { + result.max = min; + result.min = max; + } + return result; +} + +fn Range1u64 mRangeFromNIdxMCount(u64 n_idx, u64 n_count, u64 m_count) { + u64 main_idxes_per_lane = m_count / n_count; + u64 leftover_idxes_count = m_count - main_idxes_per_lane * n_count; + u64 leftover_idxes_before_this_lane_count = Min(n_idx, leftover_idxes_count); + u64 lane_base_idx = n_idx*main_idxes_per_lane + leftover_idxes_before_this_lane_count; + u64 lane_base_idx__clamped = Min(lane_base_idx, m_count); + u64 lane_opl_idx = lane_base_idx__clamped + main_idxes_per_lane + ((n_idx < leftover_idxes_count) ? 1 : 0); + u64 lane_opl_idx__clamped = Min(lane_opl_idx, m_count); + Range1u64 result = range1u64Create(lane_base_idx__clamped, lane_opl_idx__clamped); + return result; +} + +void u32Swap(u32* a, u32* b) { + int t = *a; + *a = *b; + *b = t; +} + +u32 u32ArrPartition(u32 arr[], u32 low, u32 high) { + u32 pivot = arr[high]; + u32 i = low - 1; + + for (u32 j = low; j < high; j++) { + if (arr[j] < pivot) { + i++; + u32Swap(&arr[i], &arr[j]); + } + } + u32Swap(&arr[i + 1], &arr[high]); + return i + 1; +} + +fn void u32Quicksort(u32 arr[], u32 low, u32 high) { + if (low < high) { + u32 pi = u32ArrPartition(arr, low, high); + if (pi != 0) { + u32Quicksort(arr, low, pi - 1); + } + u32Quicksort(arr, pi + 1, high); + } +} + +fn void u32ReverseArray(u32 arr[], u32 size) { + u32 start = 0; + u32 end = size - 1; + while (start < end) { + u32 temp = arr[start]; + arr[start] = arr[end]; + arr[end] = temp; + start++; + end--; + } +} + diff --git a/src/base/memory.c b/src/base/memory.c new file mode 100644 index 0000000..3d63540 --- /dev/null +++ b/src/base/memory.c @@ -0,0 +1,94 @@ +#include "all.h" + +fn u64 alignForward(u64 pointer, u64 align) { + u64 p, modulo; + assert(isPowerOfTwo(align)); + p = pointer; + // Same as (p % a) but faster as 'a' is a power of two + modulo = p & (align-1); + if (modulo != 0) { + // If 'p' address is not aligned, push the address to the + // next value which is aligned + p += align - modulo; + } + return p; +} + +fn void* arenaAlloc(Arena* arena, u64 size) { + void* memory = 0; + size = alignForward(size, DEFAULT_ALIGNMENT); + if (arena->alloc_position + size > arena->commit_position) { + if (!arena->static_size) { + u64 commit_size = size; + + commit_size += ARENA_COMMIT_SIZE - 1; + commit_size -= commit_size % ARENA_COMMIT_SIZE; + + if (arena->commit_position < arena->max) { + osMemoryCommit(arena->memory + arena->commit_position, commit_size); + arena->commit_position += commit_size; + } else { + assert(0 && "Arena is out of memory"); + } + } else { + assert(0 && "Static-Size Arena is out of memory"); + } + } + + memory = arena->memory + arena->alloc_position; + arena->alloc_position += size; + return memory; +} + +fn void* arenaAllocArraySized(Arena* arena, u64 elem_size, u64 count) { + return arenaAlloc(arena, elem_size * count); +} + +fn void arenaDealloc(Arena* arena, u64 size) { + if (size > arena->alloc_position) size = arena->alloc_position; + arena->alloc_position -= size; +} + +fn void arenaInit(Arena* arena) { + MemoryZeroStruct(arena, Arena); + arena->max = ARENA_MAX; + arena->memory = osMemoryReserve(arena->max); + arena->alloc_position = 0; + arena->commit_position = 0; + arena->static_size = false; +} + +// WARNING: segfault problems with this approach +fn void arenaInitStatic(Arena* arena, u64 max) { + MemoryZeroStruct(arena, Arena); + arena->max = max; + arena->memory = osMemoryReserve(arena->max); + osMemoryCommit(arena->memory, max + (max % ARENA_COMMIT_SIZE)); + arena->alloc_position = 0; + arena->commit_position = 0; + arena->static_size = true; +} + + +fn void arenaClear(Arena* a) { + a->alloc_position = 0; +} + +fn void arenaFree(Arena* a) { + osMemoryRelease(a->memory, a->max); +} + +ScratchMem scratchGet(void) { + ThreadContext* ctx = (ThreadContext*)osThreadContextGet(); + return tctxScratchGet(ctx); +} + +void scratchReset(ScratchMem* scratch) { + ThreadContext* ctx = (ThreadContext*)osThreadContextGet(); + tctxScratchReset(ctx, scratch); +} + +void scratchReturn(ScratchMem* scratch) { + ThreadContext* ctx = (ThreadContext*)osThreadContextGet(); + tctxScratchReturn(ctx, scratch); +} diff --git a/src/base/os.c b/src/base/os.c new file mode 100644 index 0000000..d979442 --- /dev/null +++ b/src/base/os.c @@ -0,0 +1,12 @@ +#include "all.h" + +#if OS_WINDOWS +# include "win32_os.c" +#elif OS_LINUX +# include "linux_os.c" +# include "unix_os.c" +#elif OS_MAC +# include "pthread_barrier.c" +# include "mac_os.c" +# include "unix_os.c" +#endif diff --git a/src/base/pthread_barrier.c b/src/base/pthread_barrier.c new file mode 100644 index 0000000..495cbc9 --- /dev/null +++ b/src/base/pthread_barrier.c @@ -0,0 +1,102 @@ +/* (C) Copyright 2019 Robert Sauter + * SPDX-License-Identifier: MIT + */ + +/** Pthread-barrier implementation for macOS using a pthread mutex and condition variable */ + + +#include "pthread_barrier.h" +#include + +#ifdef __APPLE__ + +int pthread_barrier_init(pthread_barrier_t *__restrict barrier, + const pthread_barrierattr_t * __restrict attr, + unsigned count) { + if (count == 0) { + return EINVAL; + } + + int ret; + + pthread_condattr_t condattr; + pthread_condattr_init(&condattr); + if (attr) { + int pshared; + ret = pthread_barrierattr_getpshared(attr, &pshared); + if (ret) { + return ret; + } + ret = pthread_condattr_setpshared(&condattr, pshared); + if (ret) { + return ret; + } + } + + ret = pthread_mutex_init(&barrier->mutex, attr); + if (ret) { + return ret; + } + + ret = pthread_cond_init(&barrier->cond, &condattr); + if (ret) { + pthread_mutex_destroy(&barrier->mutex); + return ret; + } + + barrier->count = count; + barrier->left = count; + barrier->round = 0; + + return 0; +} + +int pthread_barrier_destroy(pthread_barrier_t *barrier) { + if (barrier->count == 0) { + return EINVAL; + } + + barrier->count = 0; + int rm = pthread_mutex_destroy(&barrier->mutex); + int rc = pthread_cond_destroy(&barrier->cond); + return rm ? rm : rc; +} + + +int pthread_barrier_wait(pthread_barrier_t *barrier) { + pthread_mutex_lock(&barrier->mutex); + if (--barrier->left) { + unsigned round = barrier->round; + do { + pthread_cond_wait(&barrier->cond, &barrier->mutex); + } while (round == barrier->round); + pthread_mutex_unlock(&barrier->mutex); + return 0; + } else { + barrier->round += 1; + barrier->left = barrier->count; + pthread_cond_broadcast(&barrier->cond); + pthread_mutex_unlock(&barrier->mutex); + return PTHREAD_BARRIER_SERIAL_THREAD; + } +} + + +int pthread_barrierattr_init(pthread_barrierattr_t *attr) { + return pthread_mutexattr_init(attr); +} + +int pthread_barrierattr_destroy(pthread_barrierattr_t *attr) { + return pthread_mutexattr_destroy(attr); +} + +int pthread_barrierattr_getpshared(const pthread_barrierattr_t *__restrict attr, + int *__restrict pshared) { + return pthread_mutexattr_getpshared(attr, pshared); +} + +int pthread_barrierattr_setpshared(pthread_barrierattr_t *attr, int pshared) { + return pthread_mutexattr_setpshared(attr, pshared); +} + +#endif /* __APPLE */ \ No newline at end of file diff --git a/src/base/pthread_barrier.h b/src/base/pthread_barrier.h new file mode 100644 index 0000000..0db4acd --- /dev/null +++ b/src/base/pthread_barrier.h @@ -0,0 +1,49 @@ +/** Pthread-barrier implementation for macOS */ +#ifndef PTHREAD_BARRIER_H +#define PTHREAD_BARRIER_H + +#include + +#ifdef __APPLE__ + +#ifdef __cplusplus +extern "C" { +#endif + +#ifndef PTHREAD_BARRIER_SERIAL_THREAD +# define PTHREAD_BARRIER_SERIAL_THREAD -1 +#endif + +typedef pthread_mutexattr_t pthread_barrierattr_t; + +/* structure for internal use that should be considered opaque */ +typedef struct { + pthread_mutex_t mutex; + pthread_cond_t cond; + unsigned count; + unsigned left; + unsigned round; +} pthread_barrier_t; + +int pthread_barrier_init(pthread_barrier_t *__restrict barrier, + const pthread_barrierattr_t * __restrict attr, + unsigned count); +int pthread_barrier_destroy(pthread_barrier_t *barrier); + +int pthread_barrier_wait(pthread_barrier_t *barrier); + +int pthread_barrierattr_init(pthread_barrierattr_t *attr); +int pthread_barrierattr_destroy(pthread_barrierattr_t *attr); +int pthread_barrierattr_getpshared(const pthread_barrierattr_t *__restrict attr, + int *__restrict pshared); +int pthread_barrierattr_setpshared(pthread_barrierattr_t *attr, + int pshared); + + +#ifdef __cplusplus +} +#endif + +#endif /* __APPLE__ */ + +#endif /* PTHREAD_BARRIER_H */ diff --git a/src/base/serialize.c b/src/base/serialize.c new file mode 100644 index 0000000..ecb304d --- /dev/null +++ b/src/base/serialize.c @@ -0,0 +1,66 @@ +#include "all.h" + +fn u64 writeU64ToBufferLE(u8* buffer, u64 value) { + buffer[0] = (u8)(value & 0xFF); + buffer[1] = (u8)((value >> 8) & 0xFF); + buffer[2] = (u8)((value >> 16) & 0xFF); + buffer[3] = (u8)((value >> 24) & 0xFF); + buffer[4] = (u8)((value >> 32) & 0xFF); + buffer[5] = (u8)((value >> 40) & 0xFF); + buffer[6] = (u8)((value >> 48) & 0xFF); + buffer[7] = (u8)((value >> 56) & 0xFF); + return 8;// number of bytes written +} + +fn u64 writeU32ToBufferLE(u8* buffer, u32 value) { + buffer[0] = (u8)(value & 0xFF); + buffer[1] = (u8)((value >> 8) & 0xFF); + buffer[2] = (u8)((value >> 16) & 0xFF); + buffer[3] = (u8)((value >> 24) & 0xFF); + return 4;// number of bytes written +} + +fn u64 writeI32ToBufferLE(u8* buffer, i32 value) { + buffer[0] = (u8)(value & 0xFF); + buffer[1] = (u8)((value >> 8) & 0xFF); + buffer[2] = (u8)((value >> 16) & 0xFF); + buffer[3] = (u8)((value >> 24) & 0xFF); + return 4;// number of bytes written +} + +fn u64 writeU16ToBufferLE(u8* buffer, u16 value) { + buffer[0] = (u8)(value & 0xFF); + buffer[1] = (u8)((value >> 8) & 0xFF); + return 2;// number of bytes written +} + +fn u64 readU64FromBufferLE(u8 *buffer) { + return (u64)buffer[0] | + ((u64)buffer[1] << 8) | + ((u64)buffer[2] << 16) | + ((u64)buffer[3] << 24) | + ((u64)buffer[4] << 32) | + ((u64)buffer[5] << 40) | + ((u64)buffer[6] << 48) | + ((u64)buffer[7] << 56); +} + +fn u32 readU32FromBufferLE(u8 *buffer) { + return (u32)buffer[0] | + ((u32)buffer[1] << 8) | + ((u32)buffer[2] << 16) | + ((u32)buffer[3] << 24); +} + +fn i32 readI32FromBufferLE(u8 *buffer) { + return (i32)buffer[0] | + ((i32)buffer[1] << 8) | + ((i32)buffer[2] << 16) | + ((i32)buffer[3] << 24); +} + +fn u16 readU16FromBufferLE(u8 *buffer) { + return (u16)buffer[0] | + ((u16)buffer[1] << 8); +} + diff --git a/src/base/string.c b/src/base/string.c new file mode 100644 index 0000000..d475b7e --- /dev/null +++ b/src/base/string.c @@ -0,0 +1,216 @@ +#include "all.h" + +fn bool stringsEq(String* a, String* b) { + if (a->length != b->length) { + return false; + } + for (i32 i = 0; i < a->length; i++) { + if (a->bytes[i] != b->bytes[i]) { + return false; + } + } + return true; +} + +fn bool cStringEqString(str a, String* b) { + if (strlen(a) != b->length) { + return false; + } + for (i32 i = 0; i < b->length; i++) { + if (a[i] != b->bytes[i]) { + return false; + } + } + return true; +} + +fn Utf8Character classifyUtf8Character(u8 c) { + /*two_byte utf8 starts with 1100. 192 + three_byte utf8 starts with 1110. 224 + four_byte utf8 starts with 1111. 240*/ + if (c <= 127) { + return Utf8CharacterAscii; + } else if (c >= 192 && c < 224) { + return Utf8CharacterTwoByte; + } else if (c >= 224 && c < 240) { + return Utf8CharacterThreeByte; + } else if (c >= 240) { + return Utf8CharacterFourByte; + } else { + assert(false && "Not a valid utf8 starting byte"); + return Utf8Character_Count; + } +} + +fn bool isUtf8Ascii(u8 c) { + return classifyUtf8Character(c) == Utf8CharacterAscii; +} + +fn bool isUtf8TwoByte(u8 c) { + return classifyUtf8Character(c) == Utf8CharacterTwoByte; +} + +fn bool isUtf8ThreeByte(u8 c) { + return classifyUtf8Character(c) == Utf8CharacterThreeByte; +} + +fn bool isUtf8FourByte(u8 c) { + return classifyUtf8Character(c) == Utf8CharacterFourByte; +} + +fn u8 lowerAscii(u8 c) { + if (c >= 65 && c <= 90) { + return c + 32; + } + return c; +} + +fn u8 upperAscii(u8 c) { + if (c >= 97 && c <= 122) { + return c - 32; + } + return c; +} + +fn bool isAlphaUnderscoreSpace(u8 c) { + return ((c >= 'A' && c <= 'Z') + || (c >= 'a' && c <= 'z') + || c == ' ' + || c == '_'); +} + +fn bool isSimplePrintable(u8 c) { + return (c >= ' ' && c <= '~'); +} + +typedef struct StrDecode { + u32 codepoint; + u32 size; +} StrDecode; + +fn StrDecode strDecodeUTF8(u8 *str, u32 cap){ + u8 length[] = { + 1, 1, 1, 1, // 000xx + 1, 1, 1, 1, + 1, 1, 1, 1, + 1, 1, 1, 1, + 0, 0, 0, 0, // 100xx + 0, 0, 0, 0, + 2, 2, 2, 2, // 110xx + 3, 3, // 1110x + 4, // 11110 + 0, // 11111 + }; + u8 first_byte_mask[] = { 0, 0x7F, 0x1F, 0x0F, 0x07 }; + u8 final_shift[] = { 0, 18, 12, 6, 0 }; + + StrDecode result = {0}; + if (cap > 0){ + result.codepoint = '#'; + result.size = 1; + + u8 byte = str[0]; + u8 l = length[byte >> 3]; + if (0 < l && l <= cap){ + u32 cp = (byte & first_byte_mask[l]) << 18; + switch (l){ + case 4: cp |= ((str[3] & 0x3F) << 0); + case 3: cp |= ((str[2] & 0x3F) << 6); + case 2: cp |= ((str[1] & 0x3F) << 12); + default: break; + } + cp >>= final_shift[l]; + + result.codepoint = cp; + result.size = l; + } + } + + return result; +} + +fn u32 strEncodeUTF8(u8 *dst, u32 codepoint){ + u32 size = 0; + if (codepoint < (1 << 8)) { + dst[0] = codepoint; + size = 1; + } else if (codepoint < (1 << 11)) { + dst[0] = 0xC0 | (codepoint >> 6); + dst[1] = 0x80 | (codepoint & 0x3F); + size = 2; + } + else if (codepoint < (1 << 16)) { + dst[0] = 0xE0 | (codepoint >> 12); + dst[1] = 0x80 | ((codepoint >> 6) & 0x3F); + dst[2] = 0x80 | (codepoint & 0x3F); + size = 3; + } else if (codepoint < (1 << 21)) { + dst[0] = 0xF0 | (codepoint >> 18); + dst[1] = 0x80 | ((codepoint >> 12) & 0x3F); + dst[2] = 0x80 | ((codepoint >> 6) & 0x3F); + dst[3] = 0x80 | (codepoint & 0x3F); + size = 4; + } else { + dst[0] = '#'; + size = 1; + } + return size; +} + +fn StrDecode strDecodeUTF16(u16 *str, u32 cap){ + StrDecode result = {'#', 1}; + u16 x = str[0]; + if (x < 0xD800 || 0xDFFF < x) { + result.codepoint = x; + } else if (cap >= 2) { + u16 y = str[1]; + if (0xD800 <= x && x < 0xDC00 && + 0xDC00 <= y && y < 0xE000 + ) { + u16 xj = x - 0xD800; + u16 yj = y - 0xDc00; + u32 xy = (xj << 10) | yj; + result.codepoint = xy + 0x10000; + result.size = 2; + } + } + return result; +} + +fn u32 strEncodeUTF16(u16 *dst, u32 codepoint){ + u32 size = 0; + if (codepoint < 0x10000) { + dst[0] = codepoint; + size = 1; + } else { + u32 cpj = codepoint - 0x10000; + dst[0] = (cpj >> 10) + 0xD800; + dst[1] = (cpj & 0x3FF) + 0xDC00; + size = 2; + } + return(size); +} + +fn StringUTF16Const str16FromStr8(Arena* arena, String string) { + u16* memory = arenaAllocArray(arena, u16, string.length * 2 + 1); + + u16* dptr = memory; + u8* ptr = (u8*)string.bytes; + u8* opl = (u8*)string.bytes + string.length; + for (; ptr < opl;){ + StrDecode decode = strDecodeUTF8(ptr, (u64)(opl - ptr)); + u32 enc_size = strEncodeUTF16(dptr, decode.codepoint); + ptr += decode.size; + dptr += enc_size; + } + + *dptr = 0; + + u64 alloc_count = string.length*2 + 1; + u64 string_count = (u64)(dptr - memory); + u64 unused_count = alloc_count - string_count - 1; + arenaDealloc(arena, unused_count * sizeof(*memory)); + + StringUTF16Const result = { memory, string_count }; + return result; +} diff --git a/src/base/tctx.c b/src/base/tctx.c new file mode 100644 index 0000000..aaa9b00 --- /dev/null +++ b/src/base/tctx.c @@ -0,0 +1,83 @@ +#include "all.h" + +void tctxInit(ThreadContext* ctx) { + arenaInit(&ctx->arena); + osThreadContextSet(ctx); +} + +void tctxFree(ThreadContext* ctx) { + arenaFree(&ctx->arena); + osThreadContextSet(ctx); +} + +fn ThreadContext *tctxSelected(void) { + return (ThreadContext*)osThreadContextGet(); +} + +ScratchMem tctxScratchGet(ThreadContext* ctx) { + if (!ctx->free_list) { + ScratchMem scratch = {0}; + + scratch.arena.memory = arenaAlloc(&ctx->arena, M_SCRATCH_SIZE); + scratch.arena.max = M_SCRATCH_SIZE; + scratch.arena.alloc_position = 0; + scratch.arena.commit_position = M_SCRATCH_SIZE; + scratch.arena.static_size = true; + + ctx->max_created++; + return scratch; + } else { + ScratchMem scratch = {0}; + + scratch.arena.memory = (u8*) ctx->free_list; + scratch.arena.max = M_SCRATCH_SIZE; + scratch.arena.alloc_position = 0; + scratch.arena.commit_position = M_SCRATCH_SIZE; + scratch.arena.static_size = true; + + ctx->free_list = ctx->free_list->next; + return scratch; + } +} + +void tctxScratchReset(ThreadContext* ctx, ScratchMem* scratch) { + scratch->arena.alloc_position = 0; +} + +void tctxScratchReturn(ThreadContext* ctx, ScratchMem* scratch) { + ScratchFreeListNode* prev_head = ctx->free_list; + ctx->free_list = (ScratchFreeListNode*) scratch->arena.memory; + ctx->free_list->next = prev_head; +} + +fn LaneCtx tctxSetLaneCtx(LaneCtx lane_ctx) { + ThreadContext *tctx = tctxSelected(); + LaneCtx restore = tctx->lane_ctx; + tctx->lane_ctx = lane_ctx; + return restore; +} + +fn void tctxLaneBarrierWait(void *broadcast_ptr, u64 broadcast_size, u64 broadcast_src_lane_idx) { + ThreadContext *tctx = tctxSelected(); + + // broadcasting -> copy to broadcast memory on source lane + u64 broadcast_size_clamped = Min(broadcast_size, sizeof(tctx->lane_ctx.broadcast_memory[0])); + if(broadcast_ptr != 0 && LaneIdx() == broadcast_src_lane_idx) { + MemoryCopy(tctx->lane_ctx.broadcast_memory, broadcast_ptr, broadcast_size_clamped); + } + + // all cases: barrier + osBarrierWait(tctx->lane_ctx.barrier); + + // broadcasting -> copy from broadcast memory on destination lanes + if(broadcast_ptr != 0 && LaneIdx() != broadcast_src_lane_idx) + { + MemoryCopy(broadcast_ptr, tctx->lane_ctx.broadcast_memory, broadcast_size_clamped); + } + + // broadcasting -> barrier on all lanes + if(broadcast_ptr != 0) + { + osBarrierWait(tctx->lane_ctx.barrier); + } +} diff --git a/src/base/thread.c b/src/base/thread.c new file mode 100644 index 0000000..587a71d --- /dev/null +++ b/src/base/thread.c @@ -0,0 +1,36 @@ +#include "all.h" + +Thread spawnThread(void * (*threadFn)(void *), void* thread_arg) { + pthread_t thread; + pthread_create(&thread, NULL, threadFn, thread_arg); + Thread result = { thread }; + return result; +} + +Mutex newMutex() { + Mutex result = { PTHREAD_MUTEX_INITIALIZER }; + return result; +} + +Cond newCond() { + Cond result = { 0 }; + pthread_cond_init(&result.cond, NULL); + return result; +} + +void lockMutex(Mutex* m) { + pthread_mutex_lock(&m->mutex); +} + +void unlockMutex(Mutex* m) { + pthread_mutex_unlock(&m->mutex); +} + +void signalCond(Cond* cond) { + pthread_cond_signal(&cond->cond); +} + +void waitForCondSignal(Cond* cond, Mutex* mutex) { + pthread_cond_wait(&cond->cond, &mutex->mutex); +} + diff --git a/src/base/thread_context.h b/src/base/thread_context.h new file mode 100644 index 0000000..c18d88d --- /dev/null +++ b/src/base/thread_context.h @@ -0,0 +1,114 @@ +#include "types.h" +#include "memory.h" +#ifndef BASE_THREAD_CONTEXT_H +#define BASE_THREAD_CONTEXT_H + +typedef struct LaneCtx { + u64 lane_idx; + u64 lane_count; + Barrier barrier; + u64 *broadcast_memory; +} LaneCtx; + +typedef struct AccessPt AccessPt; +struct AccessPt +{ + u64 access_refcount; + u64 last_time_touched_us; + u64 last_update_idx_touched; +}; + +typedef struct AccessPtExpireParams AccessPtExpireParams; +struct AccessPtExpireParams +{ + u64 time; + u64 update_idxs; +}; + +typedef struct Touch Touch; +struct Touch +{ + Touch *next; + AccessPt *pt; + CondVar cv; +}; + +typedef struct Access Access; +struct Access +{ + Access *next; + Touch *top_touch; +}; + +typedef struct TCtx { + // scratch arenas + Arena *arenas[2]; + u8 thread_name[32]; + u64 thread_name_size; + + LaneCtx lane_ctx; + + // source location info + char *file_name; + u64 line_number; + + // accesses + Arena *access_arena; + Access *free_access; + Touch *free_touch; + + u64 *progress_counter_ptr; + u64 *progress_target_ptr; +} TCtx; + +//////////////////////////////// +// Thread Context Functions + +// thread-context allocation & selection +fn TCtx *tctxAlloc(void); +fn void tctxRelease(TCtx *tctx); +fn void tctxSelect(TCtx *tctx); +fn TCtx *tctxSelected(void); + +//- rjf: scratch arenas +internal Arena *tctx_get_scratch(Arena **conflicts, U64 count); +#define scratch_begin(conflicts, count) temp_begin(tctx_get_scratch((conflicts), (count))) +#define scratch_end(scratch) temp_end(scratch) + +//- rjf: lane metadata +internal LaneCtx tctx_set_lane_ctx(LaneCtx lane_ctx); +internal void tctx_lane_barrier_wait(void *broadcast_ptr, U64 broadcast_size, U64 broadcast_src_lane_idx); +#define lane_idx() (tctx_selected()->lane_ctx.lane_idx) +#define lane_count() (tctx_selected()->lane_ctx.lane_count) +#define lane_from_task_idx(idx) ((idx)%lane_count()) +#define lane_ctx(ctx) tctx_set_lane_ctx((ctx)) +#define lane_sync() tctx_lane_barrier_wait(0, 0, 0) +#define lane_sync_u64(ptr, src_lane_idx) tctx_lane_barrier_wait((ptr), sizeof(*(ptr)), (src_lane_idx)) +#define lane_range(count) m_range_from_n_idx_m_count(lane_idx(), lane_count(), (count)) + +//- rjf: thread names +internal void tctx_set_thread_name(String8 name); +internal String8 tctx_get_thread_name(void); + +//- rjf: thread source-locations +internal void tctx_write_srcloc(char *file_name, U64 line_number); +internal void tctx_read_srcloc(char **file_name, U64 *line_number); +#define tctx_write_this_srcloc() tctx_write_srcloc(__FILE__, __LINE__) + +//- rjf: access scopes +internal Access *access_open(void); +internal void access_close(Access *access); +internal void access_touch(Access *access, AccessPt *pt, CondVar cv); + +//- rjf: access points +internal B32 access_pt_is_expired_(AccessPt *pt, AccessPtExpireParams *params); +#define access_pt_is_expired(pt, ...) access_pt_is_expired_((pt), &(AccessPtExpireParams){.time = 2000000, .update_idxs = 2, __VA_ARGS__}) + +//- rjf: progress counters +#define set_progress_ptr(ptr) (tctx_selected()->progress_counter_ptr = (ptr)) +#define set_progress_target_ptr(ptr) (tctx_selected()->progress_target_ptr = (ptr)) +#define set_progress(val) (tctx_selected()->progress_counter_ptr ? ins_atomic_u64_eval_assign(tctx_selected()->progress_counter_ptr, (val)) : (void)0) +#define add_progress(val) (tctx_selected()->progress_counter_ptr ? ins_atomic_u64_add_eval(tctx_selected()->progress_counter_ptr, (val)) : (void)0) +#define set_progress_target(val) (tctx_selected()->progress_target_ptr ? ins_atomic_u64_eval_assign(tctx_selected()->progress_target_ptr, (val)) : (void)0) + +#endif // BASE_THREAD_CONTEXT_H diff --git a/src/base/unix_os.c b/src/base/unix_os.c new file mode 100644 index 0000000..c39c96f --- /dev/null +++ b/src/base/unix_os.c @@ -0,0 +1,147 @@ +#include "all.h" + +global pthread_key_t linux_thread_context_key; +// ThreadContext +void osInit() { + pthread_key_create(&linux_thread_context_key, NULL); +} + +void* osThreadContextGet() { + return pthread_getspecific(linux_thread_context_key); +} + +void osThreadContextSet(void* ctx) { + pthread_setspecific(linux_thread_context_key, ctx); +} + +bool osThreadJoin(Thread handle, u64 endt_us) { + /* + if(MemoryIsZeroStruct(&handle)) { return 0; } + OS_LNX_Entity *entity = (OS_LNX_Entity *)handle.u64[0]; + int join_result = pthread_join(entity->thread.handle, 0); + B32 result = (join_result == 0); + os_lnx_entity_release(entity); + return result; + */ + i32 join_result = pthread_join(handle.thread, NULL); + bool result = join_result == 0; + return result; +} + +// Memory +fn void* osMemoryReserve(u64 size) { + void* result = mmap(((void*)0), size, PROT_NONE, MAP_PRIVATE | MAP_ANON, -1, 0); + if(result == MAP_FAILED) { + result = 0; + } + return result; +} + +fn void osMemoryCommit(void* memory, u64 size) { + i32 result = mprotect(memory, size, PROT_READ | PROT_WRITE); + assert(result == 0 && "osMemoryCommit() failed"); +} + +fn void osMemoryDecommit(void* memory, u64 size) { + mprotect(memory, size, PROT_NONE); +} + +fn void osMemoryRelease(void* memory, u64 size) { + munmap(memory, size); +} + +// TUI +TermIOs osStartTUI(bool blocking) { + // set up the TUI incantations + printf("\033[?1049h"); // go to alternate buffer + TermIOs terminal_attributes, old_terminal_attributes; + tcgetattr(STDOUT_FILENO, &terminal_attributes); + old_terminal_attributes = terminal_attributes; + terminal_attributes.c_lflag &= ~(ICANON | ECHO); // dont echo keypresses, dont wait for carriage return + tcsetattr(STDOUT_FILENO, TCSANOW, &terminal_attributes); + if (!blocking) { + fcntl(STDIN_FILENO, F_SETFL, O_NONBLOCK); // non-blocking input mode + } + fflush(stdout); + return old_terminal_attributes; +} + +fn void osEndTUI(TermIOs old_terminal_attributes) { + tcsetattr(STDOUT_FILENO, TCSANOW, &old_terminal_attributes); + + // cleanup terminal TUI incantations + printf("\033[?1049l"); + fflush(stdout); +} + +fn Dim2 osGetTerminalDimensions() { + Dim2 result = {0}; + struct winsize ws; + if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &ws) == -1) { + //perror("ioctl TIOCGWINSZ failed"); + //exit(1); + return result; + } + result.width = ws.ws_col; + result.height = ws.ws_row; + return result; +} + +void osBlitToTerminal(ptr writeable_output_ansi_string, i64 count) { + int flags = fcntl(STDOUT_FILENO, F_GETFL); + fcntl(STDOUT_FILENO, F_SETFL, flags & ~O_NONBLOCK); + + u64 total = 0; + while (total < count) { + i64 written_bytes = write(STDOUT_FILENO, writeable_output_ansi_string + total, count - total); + /* TODO handle this error + if (written_bytes < 0) { + if (errno == EINTR) continue; // Interrupted, retry + return -1; // Real error + } + */ + total += written_bytes; + } + fcntl(STDOUT_FILENO, F_SETFL, flags); +} + +bool osInitNetwork() { return true; } + +void osReadConsoleInput(u8* buffer, u32 len) { + MemoryZero(buffer, len); // reset the input so it's not contaminated by last keystroke + read(STDIN_FILENO, buffer, len); +} + +#define LOCALHOST_127 16777343 +i32 osLanIPAddress() { // returns as HOST byte-order + i32 result = 0; + struct ifaddrs *ifaddr, *ifa; + if (getifaddrs(&ifaddr) != -1) { + for (ifa = ifaddr; ifa != NULL; ifa = ifa->ifa_next) { + if (ifa->ifa_addr == NULL) continue; + + int family = ifa->ifa_addr->sa_family; + + if (family == AF_INET) { + struct sockaddr_in addr = *(struct sockaddr_in*)ifa->ifa_addr; + if (addr.sin_addr.s_addr != LOCALHOST_127) { + freeifaddrs(ifaddr); + //printf("%s %d %d", inet_ntoa(addr.sin_addr), addr.sin_addr.s_addr, ntohl(addr.sin_addr.s_addr)); + return ntohl(addr.sin_addr.s_addr); + } + /* + char host[NI_MAXHOST]; + int s = getnameinfo(ifa->ifa_addr, sizeof(struct sockaddr_in), + host, NI_MAXHOST, NULL, 0, NI_NUMERICHOST); + + if (s == 0) { + printf("Interface: %s\tAddress: %s\n", ifa->ifa_name, host); + } + */ + } + } + + freeifaddrs(ifaddr); + } + return result; +} diff --git a/src/base/win32_os.c b/src/base/win32_os.c new file mode 100644 index 0000000..4f1b061 --- /dev/null +++ b/src/base/win32_os.c @@ -0,0 +1,326 @@ +#include "string.h" +#include "os.h" +#include +#include + +static u64 w32_ticks_per_sec = 1; +static u32 w32_thread_context_index; + +void osInit() { + LARGE_INTEGER perf_freq = {0}; + if (QueryPerformanceFrequency(&perf_freq)) { + w32_ticks_per_sec = ((u64)perf_freq.HighPart << 32) | perf_freq.LowPart; + } + timeBeginPeriod(1); + + w32_thread_context_index = TlsAlloc(); +} + +void* osThreadContextGet() { + return TlsGetValue(w32_thread_context_index); +} + +void osThreadContextSet(void* ctx) { + TlsSetValue(w32_thread_context_index, ctx); +} + +// Memory +fn void* osMemoryReserve(u64 size) { + return VirtualAlloc(0, size, MEM_RESERVE, PAGE_NOACCESS); +} + +fn void osMemoryCommit(void* memory, u64 size) { + VirtualAlloc(memory, size, MEM_COMMIT, PAGE_READWRITE); +} + +fn void osMemoryDecommit(void* memory, u64 size) { + VirtualFree(memory, size, MEM_DECOMMIT); +} + +fn void osMemoryRelease(void* memory, u64 size) { + VirtualFree(memory, 0, MEM_RELEASE); +} + +// Time +fn u64 osTimeMicrosecondsNow() { + u64 result = 0; + LARGE_INTEGER perf_counter = {0}; + if (QueryPerformanceCounter(&perf_counter)) { + u64 ticks = ((u64)perf_counter.HighPart << 32) | perf_counter.LowPart; + result = ticks * 1000000 / w32_ticks_per_sec; + } + return result; +} + +#define MICROSECONDS_PER_MILLISECOND 1000 +fn void osSleepMicroseconds(u32 t) { + Sleep(t / MICROSECONDS_PER_MILLISECOND); +} + +// Files +fn bool osFileExists(String filename) { + assert(false && "Not Implemented"); + ScratchMem scratch = scratchGet(); + StringUTF16Const filename16 = str16FromStr8(&scratch.arena, filename); + DWORD ret = GetFileAttributesW((WCHAR*)filename16.string); + scratchReturn(&scratch); + return (ret != INVALID_FILE_ATTRIBUTES && !(ret & FILE_ATTRIBUTE_DIRECTORY)); +} + +fn String osFileRead(Arena* arena, ptr filepath) { + assert(false && "Not Implemented"); + String result = {0}; + return result; +} + +fn bool osFileCreate(String filename) { + assert(false && "Not Implemented"); + return false; +} + +fn bool osFileCreateWrite(String filename, String data) { + assert(false && "Not Implemented"); + bool result = true; + return result; +} + +fn bool osFileWrite(String filename, String data) { + assert(false && "Not Implemented"); + bool result = true; + return result; +} + + +// Misc +fn void osDebugPrint(bool debug_mode, const char * format, ... ) { + if (debug_mode) { + va_list args; + va_start(args, format); + vprintf(format, args); + va_end(args); + } +} + + +// TUI +TermIOs osStartTUI(bool blocking) { + TermIOs old_settings; + + // Windows implementation + HANDLE hStdin = GetStdHandle(STD_INPUT_HANDLE); + HANDLE hStdout = GetStdHandle(STD_OUTPUT_HANDLE); + + // Save old console modes + GetConsoleMode(hStdin, &old_settings.input_mode); + GetConsoleMode(hStdout, &old_settings.output_mode); + + // Set up alternate screen buffer + printf("\033[?1049h"); + fflush(stdout); + + // Disable line input and echo (equivalent to ~ICANON and ~ECHO) + DWORD new_input_mode = old_settings.input_mode; + new_input_mode &= ~(ENABLE_LINE_INPUT | ENABLE_ECHO_INPUT); + + // Enable virtual terminal processing for ANSI escape sequences + DWORD new_output_mode = old_settings.output_mode; + new_output_mode |= ENABLE_VIRTUAL_TERMINAL_PROCESSING; + + SetConsoleMode(hStdin, new_input_mode); + SetConsoleMode(hStdout, new_output_mode); + + SetConsoleOutputCP(CP_UTF8); + + // Note: Windows console is inherently non-blocking when using + // ENABLE_LINE_INPUT disabled. You can check for input with: + // DWORD events; + // GetNumberOfConsoleInputEvents(hStdin, &events); + // Or use PeekConsoleInput() before ReadConsoleInput() + + return old_settings; +} + +fn void osEndTUI(TermIOs old_terminal_attributes) { + HANDLE hStdin = GetStdHandle(STD_INPUT_HANDLE); + HANDLE hStdout = GetStdHandle(STD_OUTPUT_HANDLE); + + // Restore console modes + SetConsoleMode(hStdin, old_terminal_attributes.input_mode); + SetConsoleMode(hStdout, old_terminal_attributes.output_mode); + + // cleanup terminal TUI incantations + printf("\033[?1049l"); + fflush(stdout); +} + +fn Dim2 osGetTerminalDimensions() { + Dim2 result = {0}; + + HANDLE hStdout = GetStdHandle(STD_OUTPUT_HANDLE); + CONSOLE_SCREEN_BUFFER_INFO csbi; + + if (GetConsoleScreenBufferInfo(hStdout, &csbi)) { + result.width = csbi.srWindow.Right - csbi.srWindow.Left + 1; + result.height = csbi.srWindow.Bottom - csbi.srWindow.Top + 1; + } else { + exit(1); + } + + return result; +} + +void osBlitToTerminal(ptr writeable_output_ansi_string, i64 count) { + HANDLE hStdout = GetStdHandle(STD_OUTPUT_HANDLE); + DWORD written; + WriteConsole(hStdout, writeable_output_ansi_string, count, &written, NULL); + assert(written == count); +} + +bool osInitNetwork() { + WSADATA wsaData; + + if (WSAStartup(MAKEWORD(2,2), &wsaData) != 0) { + return false; + } + + if (LOBYTE(wsaData.wVersion) != 2 || HIBYTE(wsaData.wVersion) != 2) { + WSACleanup(); + return false; + } + + return true; +} + +void osReadConsoleInput(u8* buffer, u32 len) { + MemoryZero(buffer, len); // reset the input so it's not contaminated by last keystroke + + if (_kbhit()) { + buffer[0] = _getch(); + bool first_byte_is_special = buffer[0] == 0 || buffer[0] == 224; + if (first_byte_is_special && _kbhit()) { + u8 windows_key = _getch(); + switch (windows_key) { + case 72: buffer[0] = 27; buffer[1] = 91; buffer[2] = 65; break; // up + case 75: buffer[0] = 27; buffer[1] = 91; buffer[2] = 68; break; // left + case 77: buffer[0] = 27; buffer[1] = 91; buffer[2] = 67; break; // right + case 80: buffer[0] = 27; buffer[1] = 91; buffer[2] = 66; break; // down + default: buffer[1] = windows_key; + } + /*if (_kbhit()) { + buffer[2] = _getch(); + if (_kbhit()) { + buffer[3] = _getch(); + } + }*/ + } + } + /* + HANDLE hStdin = GetStdHandle(STD_INPUT_HANDLE); + INPUT_RECORD ir; + DWORD read; + if (PeekConsoleInput(hStdin, &ir, 1, &read) && read > 0) { + ReadConsole(hStdin, buffer, len, &read, NULL); + ReadConsoleInput(hStdin, &ir, 1, &read); + if (ir.EventType == KEY_EVENT) { + buffer[0] = ir.Event.KeyEvent.uChar.AsciiChar; + } + } + */ +} + +i32 osLanIPAddress() { + i32 result = 0; + /* + IP_ADAPTER_INFO *pAdapterInfo; + IP_ADAPTER_INFO *pAdapter = NULL; + DWORD dwRetVal = 0; + ULONG ulOutBufLen; + + pAdapterInfo = (IP_ADAPTER_INFO *)malloc(sizeof(IP_ADAPTER_INFO)); + ulOutBufLen = sizeof(IP_ADAPTER_INFO); + + // Make an initial call to GetAdaptersInfo to get the necessary size + if (GetAdaptersInfo(pAdapterInfo, &ulOutBufLen) == ERROR_BUFFER_OVERFLOW) { + free(pAdapterInfo); + pAdapterInfo = (IP_ADAPTER_INFO *)malloc(ulOutBufLen); + } + + if ((dwRetVal = GetAdaptersInfo(pAdapterInfo, &ulOutBufLen)) == NO_ERROR) { + pAdapter = pAdapterInfo; + printf("LAN Addresses:\n"); + + // TODO: test this on windows + while (pAdapter) { + printf("Interface: %s\n", pAdapter->AdapterName); + printf("Description: %s\n", pAdapter->Description); + printf("IP Address: %s\n", pAdapter->IpAddressList.IpAddress.String); + printf("IP Address: %d\n", pAdapter->Address[0] << 24 | pAdapter->Address[1] << 16 | pAdapter->Address[2] << 8 | pAdapter->Address[3]); + printf("\n"); + pAdapter = pAdapter->Next; + } + } else { + printf("GetAdaptersInfo failed: %ld\n", dwRetVal); + } + + if (pAdapterInfo) { + free(pAdapterInfo); + } + */ + WSADATA wsa; + char hostname[256]; + struct addrinfo hints, *final = NULL, *ptr = NULL; + struct sockaddr_in *sockaddr_ipv4; + + // Initialize Winsock + if (WSAStartup(MAKEWORD(2, 2), &wsa) != 0) { + return 0; + } + + // Get hostname + if (gethostname(hostname, sizeof(hostname)) == SOCKET_ERROR) { + WSACleanup(); + return 0; + } + + // Set up hints for getaddrinfo + ZeroMemory(&hints, sizeof(hints)); + hints.ai_family = AF_INET; // IPv4 + hints.ai_socktype = SOCK_STREAM; + hints.ai_protocol = IPPROTO_TCP; + + // Get address info + if (getaddrinfo(hostname, NULL, &hints, &final) != 0) { + WSACleanup(); + return 0; + } + + // Loop through results to find a non-loopback address + for (ptr = final; ptr != NULL; ptr = ptr->ai_next) { + sockaddr_ipv4 = (struct sockaddr_in *)ptr->ai_addr; + unsigned long addr = ntohl(sockaddr_ipv4->sin_addr.s_addr); + + // Skip loopback addresses (127.x.x.x) + if ((addr >> 24) != 127) { + result = sockaddr_ipv4->sin_addr.s_addr; + break; + } + } + + freeaddrinfo(final); + WSACleanup(); + + return ntohl(result); +} + +bool osThreadJoin(Thread handle, u64 endt_us) { + DWORD sleep_ms = os_w32_sleep_ms_from_endt_us(endt_us); + OS_W32_Entity *entity = (OS_W32_Entity *)PtrFromInt(handle.u64[0]); + DWORD wait_result = WAIT_OBJECT_0; + if(entity != 0) + { + wait_result = WaitForSingleObject(entity->thread.handle, sleep_ms); + CloseHandle(entity->thread.handle); + os_w32_entity_release(entity); + } + return (wait_result == WAIT_OBJECT_0); +} diff --git a/src/client.c b/src/client.c new file mode 100644 index 0000000..cb0ca33 --- /dev/null +++ b/src/client.c @@ -0,0 +1,743 @@ +/* + * Code conventions: + * MyStructType + * myFunction() + * MyMacro() + * my_variable + * MY_CONSTANT + * */ +#include +#include "base/impl.c" +#include "lib/network.c" +#include "render.c" +#include "string_chunk.c" +//#include "assets/asset1.h" +//#include "assets/asset2.h" +//#include "assets/asset3.h" + +///// #define a bunch of client-only tunable game constants +#define SYSTEM_MESSAGES_LEN 32 +#define MAX_SYSTEM_MESSAGE_LEN 512 +#define GOAL_LOOPS_PER_S 50 +#define GOAL_LOOP_US 1000000/GOAL_LOOPS_PER_S +#define LOGIN_NAME_BUFFER_LEN 16 +#define PARSED_SERVER_MESSAGE_THREAD_QUEUE_LEN 16 +#define MAIN_GAME_TAB_COUNT (2) + +///// TYPES +typedef enum Screen { + ScreenLogin, + ScreenCreateCharacter, + ScreenMainGame, + ScreenDefeat, + ScreenVictory, + Screen_Count +} Screen; + +typedef enum LoginScreenState { + LoginScreenStateInit, + LoginScreenStateLoading, + LoginScreenStateBadPw, + LoginScreenState_Count +} LoginScreenState; + +typedef enum SpeechTailDirection { + SpeechTailDirectionBottomRight, + SpeechTailDirectionBottomLeft, + SpeechTailDirectionRight, + SpeechTailDirectionLeft, + SpeechTailDirection_Count +} SpeechTailDirection; + +typedef struct Entity { + EntityType type; + u8 x; + u8 y; + u8 color; + u64 features; + u64 id; + StringChunkList name; +} Entity; + +typedef struct EntityList { + u64 length; // the currently used length + u64 capacity; + Entity* items; +} EntityList; + +typedef struct ParsedServerMessage { + Message type; + u16 port; + u16 port2; + u32 ip; + u32 ip2; + u64 id; + u64 server_frame; + XYZ xyz; + //Entity entities[PARSED_CLIENT_ENTITY_LEN]; + //u64 ids[PARSED_IDS_LEN]; +} ParsedServerMessage; + +typedef struct ParsedServerMessageThreadQueue { + ParsedServerMessage items[PARSED_SERVER_MESSAGE_THREAD_QUEUE_LEN]; + u32 head; + u32 tail; + u32 count; + Mutex mutex; + Cond not_empty; + Cond not_full; +} ParsedServerMessageThreadQueue; + +typedef struct LoginState { + LoginScreenState state; + u8 selected_field; + u8 field_index; + String name; + String password; +} LoginState; + +typedef struct MenuState { + u8 selected_index; + u8 len; + u64 id; +} MenuState; + +typedef struct GameState { + Screen screen; + Screen old_screen; + EntityList entities; + Entity me; + u64 server_frame; + u64 loop_count; + Arena entity_arena; + StringArena string_arena; + LoginState login_state; + MenuState menu; + MenuState section; // for tabbing through selected "portions" of the screen + u8 choices[4]; + UDPMessage keep_alive_msg; + UDPClient client; + StringChunkList message_input; +} GameState; + +///// GLOBALS +global bool should_quit; +global bool debug_mode = false; +global Arena permanent_arena = {0}; +global u8List system_messages[SYSTEM_MESSAGES_LEN] = {0}; +global u8 system_message_index = 0; +global GameState state = {0}; +global OutgoingMessageQueue* network_send_queue = {0}; +global ParsedServerMessageThreadQueue* network_recv_queue = {0}; +global str TABS[] = {"Debug", "Speak"}; + +///// FUNCTIONS +fn ParsedServerMessageThreadQueue* newPSMThreadQueue(Arena* a) { + ParsedServerMessageThreadQueue* result = arenaAlloc(a, sizeof(ParsedServerMessageThreadQueue)); + MemoryZero(result, (sizeof *result)); + result->mutex = newMutex(); + result->not_full = newCond(); + result->not_empty = newCond(); + return result; +} + +fn void psmThreadSafeQueuePush(ParsedServerMessageThreadQueue* queue, ParsedServerMessage* msg) { + lockMutex(&queue->mutex); { + while (queue->count == PARSED_SERVER_MESSAGE_THREAD_QUEUE_LEN) { + waitForCondSignal(&queue->not_full, &queue->mutex); + } + + MemoryCopy(&queue->items[queue->tail], msg, (sizeof *msg)); + queue->tail = (queue->tail + 1) % PARSED_SERVER_MESSAGE_THREAD_QUEUE_LEN; + queue->count++; + + signalCond(&queue->not_empty); + } unlockMutex(&queue->mutex); +} + +fn ParsedServerMessage* psmThreadSafeNonblockingQueuePop(ParsedServerMessageThreadQueue* q, ParsedServerMessage* copy_target) { + // immediately returns NULL if there's nothing in the ThreadQueue + // copies the ParsedServerMessage into `copy_target` if there is something in the queue + // and marks it as popped from the queue + ParsedServerMessage* result = NULL; + + lockMutex(&q->mutex); { + if (q->count > 0) { + result = &q->items[q->head]; + MemoryCopy(copy_target, result, (sizeof *copy_target)); + q->head = (q->head + 1) % PARSED_SERVER_MESSAGE_THREAD_QUEUE_LEN; + q->count--; + + signalCond(&q->not_full); + } + } unlockMutex(&q->mutex); + + return result; +} + +fn void entityPush(EntityList* list, Entity e) { + if (list->length >= list->capacity) { + arenaAllocArray(&state.entity_arena, Entity, list->capacity); + list->capacity = list->capacity * 2; + } + list->items[list->length] = e; + list->length += 1; +} + +fn bool entityDelete(EntityList* list, u64 id) { + assert(list->length > 0); + if (list->items[list->length - 1].id == id) { + list->length -= 1; + return true; + } else { + for (u32 i = 0; i < list->length; i++) { + if (list->items[i].id == id) { + // copy the last one over the one we are deleting + list->items[i] = list->items[list->length - 1]; + list->length -= 1; + return true; + } + } + } + return false; +} + +fn void addSystemMessage(u8* msg) { + // save the message to our system_messages ring buffer + memset(system_messages[system_message_index].items, 0, SYSTEM_MESSAGES_LEN); + sprintf((char*)system_messages[system_message_index].items, "%s", msg); + system_messages[system_message_index].length = strlen((char*)system_messages[system_message_index].items); + system_message_index += 1; + if (system_message_index == SYSTEM_MESSAGES_LEN) { + system_message_index = 0; + } +} + +fn void renderSystemMessages(Pixel* buf, Dim2 screen_dimensions, Box sys_msg_box) { + i32 printable_lines = sys_msg_box.height - 2; + if (printable_lines > SYSTEM_MESSAGES_LEN) { + printable_lines = SYSTEM_MESSAGES_LEN; + } + for (i32 i = 0; i < printable_lines; i++) { + i32 index = (system_message_index - 1 - i); + if (index < 0) { + index = SYSTEM_MESSAGES_LEN + index; + } + u32 y = sys_msg_box.y + (sys_msg_box.height - i) - 1; + u8List sys_msg = system_messages[index]; + for (i32 j = 0; j < MAX_SYSTEM_MESSAGE_LEN && j < sys_msg_box.width-4; j++) { + u32 pos = (sys_msg_box.x + 2+j) + (screen_dimensions.width * y); + if (j < sys_msg.length) { + if (sys_msg.items[j] != '\n') { + buf[pos].bytes[0] = sys_msg.items[j]; + } + } + } + } +} + +fn void renderPercentBar(TuiState* tui, u16 x, u16 y, u16 width, u8 ansi_color, u64 value, u64 max) { + Pixel* buf = tui->frame_buffer; + Dim2 screen_dimensions = tui->screen_dimensions; + u16 pos = x + (screen_dimensions.width * y); + buf[pos].bytes[0] = '['; + pos = x+width + (screen_dimensions.width * y); + buf[pos].bytes[0] = ']'; + pos = x+1 + (screen_dimensions.width * y); + f32 base_ratio = 0; + if (max != 0) { + base_ratio = ((f32)value / (f32)max); + } + f32 raw_ratio = base_ratio * (width-2); + u32 full_spaces_count = (u32) raw_ratio; + f32 remainder = raw_ratio - full_spaces_count; + for (u32 i = 0; i < full_spaces_count; i++) { + buf[pos+i].foreground = ansi_color; + renderUtf8CharToBuffer(buf, x+1+i, y, "█", screen_dimensions); + } + buf[pos+full_spaces_count].foreground = ansi_color; + if (value == max) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "█", screen_dimensions); + } else if (remainder > 0.875) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "▉", screen_dimensions); + } else if (remainder > 0.75) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "▊", screen_dimensions); + } else if (remainder > 0.625) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "▋", screen_dimensions); + } else if (remainder > 0.5) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "▌", screen_dimensions); + } else if (remainder > 0.375) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "▍", screen_dimensions); + } else if (remainder > 0.25) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "▎", screen_dimensions); + } else if (remainder > 0.125) { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, "▏", screen_dimensions); + } else { + renderUtf8CharToBuffer(buf, x+1+full_spaces_count, y, " ", screen_dimensions); + } +} + +fn void renderSpeechBubble(TuiState* tui, u16 x, u16 y, u16 max_width, String message, SpeechTailDirection dir) { + Pixel* buf = tui->frame_buffer; + Dim2 screen_dimensions = tui->screen_dimensions; + // calculate dimenions + u32 height = (message.length / max_width) + 1; + u32 width = message.length; + if (message.length > max_width) { + width = max_width; + } + if (width < 12) { + width = 12; // minimum width + } + + // print the upper border + renderUtf8CharToBuffer(buf, x, y, "╭", screen_dimensions); + for (i32 i = 0; i < width+1; i++) { + renderUtf8CharToBuffer(buf, x+1+i, y, "─", screen_dimensions); + } + renderUtf8CharToBuffer(buf, x+width+1, y, "╮", screen_dimensions); + + // start printing the rows + for (u32 i = 0; i < height; i++) { + renderUtf8CharToBuffer(buf, x, y+1+i, "│", screen_dimensions); + // print the line of text + for (u32 j = 0; j < width; j++) { + u32 character_index = (i*(width-1))+j; + u16 pos = (x+1+j) + (screen_dimensions.width * (y+1+i)); + if (message.length > character_index) { + buf[pos].bytes[0] = message.bytes[character_index]; + } else { + buf[pos].bytes[0] = ' '; + } + } + renderUtf8CharToBuffer(buf, x+1+width, y+1+i, "│", screen_dimensions); + } + + // print the bottom box border + renderUtf8CharToBuffer(buf, x, y+1+height, "╰", screen_dimensions); + for (i32 i = 0; i < width+1; i++) { + renderUtf8CharToBuffer(buf, x+1+i, y+1+height, "─", screen_dimensions); + } + renderUtf8CharToBuffer(buf, x+width+1, y+1+height, "╯", screen_dimensions); + // TODO actually print the speech tail +} + +fn void renderStaticAssetToPixelBuffer(TuiState* tui, u8* asset, u32 len, u16 x, u16 y) { + u16 line = 0; + u16 x_in_line = 0; + u16 pos = x + (tui->screen_dimensions.width * y); + for (u32 i = 0; i < len; i++, x_in_line++) { + if (asset[i] == '\n') { + line += 1; + x_in_line = 0; + pos = x + (tui->screen_dimensions.width * (y+line)); + } else if (asset[i] == ' ') { + // do nothing, we skip spaces in our assets + } else { + tui->frame_buffer[pos+x_in_line].bytes[0] = asset[i]; + } + } +} + +fn void clearServerSentState() { + //memset(&state.current_room, 0, sizeof(RenderableRoom)); + + arenaClear(&state.entity_arena); + state.entities.capacity = 64; + state.entities.length = 0; + state.entities.items = arenaAllocArray(&state.entity_arena, Entity, state.entities.capacity); +} + +fn void handleIncomingMessage(u8* message, u32 len, SocketAddress sender, i32 socket) { + Message msg_type = message[0]; + dbg("handleIncomingMessage() of len=%d, message=%s\n", len, MESSAGE_STRINGS[msg_type]); + u8List bytes = {len, len, message}; + u64 msg_pos = 1; + ParsedServerMessage parsed = {0}; + parsed.type = msg_type; + switch (msg_type) { + case MessageNewAccountCreated: + case MessageBadPw: {/*nothing to parse but the type*/} break; + case MessageCharacterId: { + parsed.id = readU64FromBufferLE(message + 1); + } break; + case Message_Count: + assert(false && "invalid msg_type detected"); + break; + case MessageInvalid: + addSystemMessage((u8*)"NOT IMPLEMENTED"); + break; + } + psmThreadSafeQueuePush(network_recv_queue, &parsed); +} + +fn void* receiveNetworkUpdates(void* udp) { + UDPClient client = *(UDPClient*)udp; + dbg("receiveNetworkUpdates() sock=%d\n", client.socket); + UDPServer server = { + .ready = true, + .server_address = client.server_address, + .server_socket = client.socket + }; + infiniteReadUDPServer(&server, handleIncomingMessage); + return NULL; +} + +fn void* sendNetworkUpdates(void* udp) { + i32 socket_fd = ((UDPClient*)udp)->socket; + dbg("sendNetworkUpdates() sock=%d\n", socket_fd); + u8List bytes_list = {0}; + while (!should_quit) { + UDPMessage msg = {0}; + outgoingMessageQueuePop(network_send_queue, &msg); + bytes_list.items = msg.bytes; + bytes_list.length = msg.bytes_len; + bytes_list.capacity = msg.bytes_len; + sendUDPu8List(socket_fd, &msg.address, &bytes_list); + } + return NULL; +} + +fn bool updateAndRender(TuiState* tui, void* s, u8* input_buffer, u64 loop_count) { + GameState* state = (GameState*) s; + state->loop_count = loop_count; + UDPClient* udp = &state->client; + state->old_screen = state->screen; + Dim2 screen_dimensions = tui->screen_dimensions; + bool game_screen_changed = state->old_screen != state->screen; // detect new screens + tui->redraw = tui->redraw || game_screen_changed; + if (screen_dimensions.height > MAX_SCREEN_HEIGHT || screen_dimensions.width > MAX_SCREEN_WIDTH) { + printf("\033[2J\033[%d;%df Your screen is too damn big. Shrink it to %dx%d max, bro... geez", 1,1, MAX_SCREEN_WIDTH, MAX_SCREEN_HEIGHT); + fflush(stdout); + return should_quit; + } + + // process server messages + u32 msg_iters = 0; + ParsedServerMessage msg = {0}; + ParsedServerMessage* next_net_msg = psmThreadSafeNonblockingQueuePop(network_recv_queue, &msg); + while (next_net_msg != NULL) { + msg_iters += 0; + switch (msg.type) { + case MessageNewAccountCreated: { + state->screen = ScreenCreateCharacter; + } break; + case MessageBadPw: { + state->login_state.state = LoginScreenStateBadPw; + } break; + case MessageCharacterId: { + state->me.id = msg.id; + char bytes[256] = {0}; + sprintf(bytes,"got character id: %lld", msg.id); + addSystemMessage((u8*)bytes); + state->screen = ScreenMainGame; + state->menu.selected_index = 0; + state->section.selected_index = 0; + } break; + case Message_Count: + case MessageInvalid: + assert(false && "invalid message from queue"); + break; + } + next_net_msg = psmThreadSafeNonblockingQueuePop(network_recv_queue, &msg); + msg_iters++; + } + + // operate on screen-based input+state + bool user_pressed_esc = input_buffer[0] == ASCII_ESCAPE && input_buffer[1] == 0; + bool user_pressed_a_number = input_buffer[0] >= '1' && input_buffer[0] <= '9' && input_buffer[1] == 0; + bool user_pressed_up = input_buffer[0] == 27 && input_buffer[1] == 91 && input_buffer[2] == 65; + bool user_pressed_down = input_buffer[0] == 27 && input_buffer[1] == 91 && input_buffer[2] == 66; + bool user_pressed_left = input_buffer[0] == 27 && input_buffer[1] == 91 && input_buffer[2] == 68; + bool user_pressed_right = input_buffer[0] == 27 && input_buffer[1] == 91 && input_buffer[2] == 67; + bool user_pressed_backspace = input_buffer[0] == ASCII_BACKSPACE || input_buffer[0] == ASCII_DEL; + bool user_pressed_enter = input_buffer[0] == ASCII_RETURN || input_buffer[0] == ASCII_LINE_FEED; + switch (state->screen) { + case ScreenCreateCharacter: { + //// SIMULATION + state->screen = ScreenMainGame; // TODO change this if you want to actually have character creation + if (state->me.id != 0) { + state->screen = ScreenMainGame; + break; + } + + if (input_buffer[0] == 'q' || user_pressed_esc) { + should_quit = true; + } else if (user_pressed_down || user_pressed_right) { + state->menu.selected_index++; + } else if (user_pressed_up || user_pressed_left) { + state->menu.selected_index--; + } else if (user_pressed_a_number) { + state->menu.selected_index = input_buffer[0] - '1'; + } else if (input_buffer[0] == ASCII_RETURN || input_buffer[0] == ASCII_LINE_FEED) { + if (state->section.selected_index == 0) { + state->choices[0] = state->menu.selected_index; + state->section.selected_index += 1; + state->menu.selected_index = 0; + state->menu.len = 2; + } else { + // send character color to server + UDPMessage msg = {0}; + msg.address = udp->server_address; + msg.bytes_len = 2; + // 1. msg type/CommandType + msg.bytes[0] = CommandCreateCharacter; + // 2. what color + msg.bytes[1] = ANSI_HP_RED;//WIZARD_COLORS[state->choices[0]]; + + outgoingMessageQueuePush(network_send_queue, &msg); + + state->screen = ScreenMainGame; + } + } + + if (state->menu.selected_index >= state->menu.len) { + state->menu.selected_index = 0; + } + + //// RENDERING + u16 line = 2; + // 1. draw the outline + Box b = { .x = 2, .y = line, .width = screen_dimensions.width - 5, .height = screen_dimensions.height - 5 }; + drawAnsiBox(tui->frame_buffer, b, screen_dimensions, true); + + // 2. draw the header label + str label = "Create Character"; + renderStrToBuffer(tui->frame_buffer, (screen_dimensions.width - (strlen(label)+4))/2, ++line, label, screen_dimensions); + + line++; + // TODO: game-specific character creation stuff + } break; + case ScreenMainGame: { + // SIMULATION + if (input_buffer[0] == 'q' || user_pressed_esc) { + should_quit = true; + } + + // RENDERING + renderStrToBuffer(tui->frame_buffer, 5, 1, "Games typically would render something here...", screen_dimensions); + bool room_active = state->section.selected_index == 0; + u32 tabs_y = 1 + 2 + 2; + u32 tx = 2; + for (u32 i = 0; i < MAIN_GAME_TAB_COUNT; i++) { + u32 tab_len = strlen(TABS[i]); + Box b = { .x = tx, .y = tabs_y, .width = tab_len+1, .height = 1 }; + drawAnsiBox(tui->frame_buffer, b, screen_dimensions, state->menu.selected_index == i); + renderStrToBuffer(tui->frame_buffer, tx+1, tabs_y+ 1, TABS[i], screen_dimensions); + tx += (tab_len + 3); + } + // draw the system messages box + u32 box_y = tabs_y + 3; + Box box = { + .x = 1, + .y = box_y, + .width = screen_dimensions.width - 4, + .height = screen_dimensions.height - box_y - 2, + }; + drawAnsiBox(tui->frame_buffer, box, screen_dimensions, !room_active); + if (state->menu.selected_index == 0) { // 0 is TABS[0] is "Debug" + renderSystemMessages(tui->frame_buffer, tui->screen_dimensions, box); + } else if (state->menu.selected_index == 1) { // 1 is TABS[1] is "Speak" + renderStrToBuffer(tui->frame_buffer, box.x+2, box.y+1, "You haven't heard anything interesting lately...", screen_dimensions); + } + } break; + case ScreenLogin: { + // SIMULATION + if (user_pressed_esc) { + should_quit = true; + } + if (state->login_state.state == LoginScreenStateLoading) { + break; + } + String* field; + if (state->login_state.selected_field == 0) { + field = &state->login_state.name; + } else { + field = &state->login_state.password; + } + if (isAlphaUnderscoreSpace(input_buffer[0])) { + field->bytes[state->login_state.field_index] = input_buffer[0]; + if (state->login_state.field_index+1 < field->capacity) { + state->login_state.field_index += 1; + field->length += 1; + } + } else if (user_pressed_backspace) { + field->bytes[state->login_state.field_index] = 0; + if (state->login_state.field_index > 0) { + state->login_state.field_index -= 1; + field->length -= 1; + } + } else if (input_buffer[0] == ASCII_RETURN || input_buffer[0] == ASCII_LINE_FEED) { + if (state->login_state.selected_field == 0) { + state->login_state.selected_field = 1; + state->login_state.field_index = 0; + } else { + u32 msg_idx = 0; + // drop the login message into the network_send_queue + UDPMessage msg = {0}; + msg.address = udp->server_address; + // 1. msg type/CommandType + msg.bytes[msg_idx++] = CommandLogin; + // 2. our LAN-IP to handle the case where we are on the same LAN as the guy we are trying to fight + // and our "listened" UDP port + msg_idx += writeU16ToBufferLE(msg.bytes + msg_idx, ~(udp->client_port)); + msg_idx += writeI32ToBufferLE(msg.bytes + msg_idx, ~osLanIPAddress()); + // 2. how long is the name + msg.bytes[msg_idx++] = state->login_state.name.length; + // 3. the name + memcpy(msg.bytes + msg_idx, state->login_state.name.bytes, state->login_state.name.length); + // 4. the password + memcpy(msg.bytes + msg_idx + state->login_state.name.length, state->login_state.password.bytes, state->login_state.password.length); + msg.bytes_len = msg_idx + state->login_state.name.length + state->login_state.password.length; + addSystemMessage((u8*)state->login_state.name.bytes); + addSystemMessage((u8*)state->login_state.password.bytes); + outgoingMessageQueuePush(network_send_queue, &msg); + // show loading signal, stop accepting input + state->login_state.state = LoginScreenStateLoading; + } + } + + // RENDERING + // TODO: 1. DiffieHelman exchange + // 2. encrypt password + // 3. send [CommandLogin uname-len username encrypted_pass] + if (screen_dimensions.height < 30 || screen_dimensions.width < 80) { + str warning_label = "Screen too smol. Plz resize."; + renderStrToBuffer(tui->frame_buffer, 6, 2, warning_label, screen_dimensions); + } else { + u32 width = screen_dimensions.width / 3; + u32 height = screen_dimensions.height / 4; + u32 x = (screen_dimensions.width - width) / 2; // center + u32 y = (screen_dimensions.height - height) / 2; // center + // outline + Box b = { .x = x, .y = y, .width = width, .height = height }; + drawAnsiBox(tui->frame_buffer, b, screen_dimensions, true); + // labels + str login_label = " Please login:"; + renderStrToBuffer(tui->frame_buffer, x+1, y+1, login_label, screen_dimensions); + str name_label = " Name: "; + renderStrToBuffer(tui->frame_buffer, x+1, y+2, name_label, screen_dimensions); + str pw_label = " Password: "; + renderStrToBuffer(tui->frame_buffer, x+1, y+3, pw_label, screen_dimensions); + // render name + u16 pos = ((y+2)*screen_dimensions.width) + (x+11); + for (u32 i = 0; i < state->login_state.name.length; i++) { + tui->frame_buffer[pos+i].bytes[0] = state->login_state.name.bytes[i]; + } + // render password + pos = ((y+3)*screen_dimensions.width) + (x+15); + for (u32 i = 0; i < state->login_state.password.length; i++) { + tui->frame_buffer[pos+i].bytes[0] = '*'; + } + if (state->login_state.state == LoginScreenStateLoading) { + renderStrToBuffer(tui->frame_buffer, x+5, y+5, "Loading...", screen_dimensions); + } else if (state->login_state.state == LoginScreenStateBadPw) { + renderStrToBuffer(tui->frame_buffer, x+5, y+5, "Bad password, dumbass", screen_dimensions); + } + // calculate where to put the cursor + if (state->login_state.selected_field == 0) { + tui->cursor.y = y+2; + tui->cursor.x = x+11+state->login_state.name.length; + } else { + tui->cursor.y = y+3; + tui->cursor.x = x+15+state->login_state.password.length; + } + } + } break; + case ScreenVictory: { + // SIMULATION + if (user_pressed_esc || input_buffer[0] == ASCII_RETURN || input_buffer[0] == ASCII_LINE_FEED) { + state->screen = ScreenMainGame; + } + + // RENDERING + renderStrToBuffer(tui->frame_buffer, 10, 10, "You WIN!!!", screen_dimensions); + } break; + case ScreenDefeat: { + // SIMULATION + if (user_pressed_esc || input_buffer[0] == ASCII_RETURN || input_buffer[0] == ASCII_LINE_FEED) { + state->me.id = 0; + state->menu.selected_index = 0; + state->section.selected_index = 0; + clearServerSentState(); + state->screen = ScreenCreateCharacter; + } + + // RENDERING + renderStrToBuffer(tui->frame_buffer, 10, 10, "You died...", screen_dimensions); + renderStrToBuffer(tui->frame_buffer, 10, 11, "[press ESC to continue]", screen_dimensions); + } break; + case Screen_Count: + assert(false && "unhandled screen"); + break; + } + + if (loop_count % 10 == 0) { + outgoingMessageQueuePush(network_send_queue, &state->keep_alive_msg); + //outgoingMessageQueuePush(network_send_queue, &testm); + } + + return should_quit; +} + +i32 main(i32 argc, ptr argv[]) { + assert(Message_Count < 256); + assert(CommandType_Count < 256); + osInit(); + // 3 threads, each their own loop: + // 1. recvNetwork() + // 2. sendNetwork() + // 3. input/update/draw "normal" gameloop + // - which is actually "wide" doing each room in parallel across N threads + + // clear and init the state + arenaInit(&permanent_arena); + arenaInit(&state.entity_arena); + state.entities.capacity = 64; + state.entities.length = 0; + state.entities.items = arenaAllocArray(&state.entity_arena, Entity, state.entities.capacity); + arenaInit(&state.string_arena.a); + state.string_arena.mutex = newMutex(); + state.message_input = stringChunkListInit(&state.string_arena); + for (i32 i = 0; i < SYSTEM_MESSAGES_LEN; i++) { + system_messages[i].capacity = MAX_SYSTEM_MESSAGE_LEN; + system_messages[i].length = 0; + system_messages[i].items = arenaAllocArraySized(&permanent_arena, sizeof(u8), MAX_SYSTEM_MESSAGE_LEN); + } + + // network queues + network_send_queue = newOutgoingMessageQueue(&permanent_arena); + network_recv_queue = newPSMThreadQueue(&permanent_arena); + + // draw login screen, get input username + password + state.screen = ScreenLogin; + state.login_state.name.capacity = LOGIN_NAME_BUFFER_LEN; + state.login_state.name.bytes = arenaAllocArray(&permanent_arena, char, state.login_state.name.capacity); + state.login_state.password.capacity = LOGIN_NAME_BUFFER_LEN; + state.login_state.password.bytes = arenaAllocArray(&permanent_arena, char, state.login_state.password.capacity); + + // network incantation for connecting to the server + if (osInitNetwork() != true) { + printf("bad network startup"); + exit(1); + } + state.client = createUDPClient(7777, argc > 1 ? argv[1] : NULL); + + // "hardcoded" keep alive message to periodically send to server + state.keep_alive_msg.address = state.client.server_address; + state.keep_alive_msg.bytes[0] = CommandKeepAlive; + state.keep_alive_msg.bytes_len = 1; + + Thread recv_thread = spawnThread(&receiveNetworkUpdates, &state.client); + Thread send_thread = spawnThread(&sendNetworkUpdates, &state.client); + + // game loop (read input, simulate next frame, render) + infiniteUILoop( + MAX_SCREEN_WIDTH, + MAX_SCREEN_HEIGHT, + GOAL_LOOP_US, + &state, + updateAndRender + ); + return 0; +} diff --git a/src/lib/network.c b/src/lib/network.c new file mode 100644 index 0000000..9278e89 --- /dev/null +++ b/src/lib/network.c @@ -0,0 +1,198 @@ +#include "../base/all.h" + +// https://stackoverflow.com/questions/1098897/what-is-the-largest-safe-udp-packet-size-on-the-internet +#define UDP_MAX_MESSAGE_LEN 508 +#ifndef NET_OUTGOING_MESSAGE_QUEUE_LEN +#define NET_OUTGOING_MESSAGE_QUEUE_LEN 16 +#endif + +typedef struct sockaddr_in SocketAddress; + +typedef struct UDPServer { + bool ready; + SocketAddress server_address; + i32 server_socket; +} UDPServer; + +typedef struct UDPClient { + bool ready; + SocketAddress server_address; + u16 client_port; + i32 socket; +} UDPClient; + +typedef struct UDPMessage { + u16 bytes_len; + SocketAddress address; + u8 bytes[UDP_MAX_MESSAGE_LEN]; +} UDPMessage; + +typedef struct OutgoingMessageQueue { + UDPMessage items[NET_OUTGOING_MESSAGE_QUEUE_LEN]; + u32 head; + u32 tail; + u32 count; + Mutex mutex; + Cond not_empty; + Cond not_full; +} OutgoingMessageQueue; + +fn OutgoingMessageQueue* newOutgoingMessageQueue(Arena* a) { + OutgoingMessageQueue* result = arenaAlloc(a, sizeof(OutgoingMessageQueue)); + MemoryZero(result, (sizeof *result)); + result->mutex = newMutex(); + result->not_full = newCond(); + result->not_empty = newCond(); + return result; +} + +fn void outgoingMessageQueuePush(OutgoingMessageQueue* queue, UDPMessage* msg) { + lockMutex(&queue->mutex); { + while (queue->count == NET_OUTGOING_MESSAGE_QUEUE_LEN) { + waitForCondSignal(&queue->not_full, &queue->mutex); + } + + MemoryCopy(&queue->items[queue->tail], msg, (sizeof *msg)); + queue->tail = (queue->tail + 1) % NET_OUTGOING_MESSAGE_QUEUE_LEN; + queue->count++; + + signalCond(&queue->not_empty); + } unlockMutex(&queue->mutex); +} + +fn UDPMessage* outgoingMessageNonblockingQueuePop(OutgoingMessageQueue* q, UDPMessage* copy_target) { + // immediately returns NULL if there's nothing in the ThreadQueue + // copies the ParsedClientCommand into `copy_target` if there is something in the queue + // and marks it as popped from the queue + UDPMessage* result = NULL; + + lockMutex(&q->mutex); { + if (q->count > 0) { + result = &q->items[q->head]; + MemoryCopy(copy_target, result, (sizeof *copy_target)); + q->head = (q->head + 1) % NET_OUTGOING_MESSAGE_QUEUE_LEN; + q->count--; + + signalCond(&q->not_full); + } + } unlockMutex(&q->mutex); + + return result; +} + +fn UDPMessage* outgoingMessageQueuePop(OutgoingMessageQueue* q, UDPMessage* copy_target) { + UDPMessage* result = NULL; + + lockMutex(&q->mutex); { + while (q->count == 0) { + waitForCondSignal(&q->not_empty, &q->mutex); + } + + result = &q->items[q->head]; + MemoryCopy(copy_target, result, (sizeof *copy_target)); + q->head = (q->head + 1) % NET_OUTGOING_MESSAGE_QUEUE_LEN; + q->count--; + + signalCond(&q->not_full); + } unlockMutex(&q->mutex); + + return result; +} + + +fn bool socketAddressEqual(SocketAddress a, SocketAddress b) { + return a.sin_addr.s_addr == b.sin_addr.s_addr + && a.sin_port == b.sin_port; +} + +// ONLY WORKS ON POSIX. taken from https://gist.github.com/miekg/a61d55a8ec6560ad6c4a2747b21e6128 + +// the only real difference between a udp "server" and a "client" is the bind() syscall +// that the server makes in order to specify a port/address that it's listening on +UDPServer createUDPServer(u16 server_port) { + UDPServer result = {0}; + // define the address we'll be listening on + result.server_address.sin_family = AF_INET; + result.server_address.sin_addr.s_addr = inet_addr("0.0.0.0");//htonl(INADDR_ANY); + result.server_address.sin_port = htons(server_port); + + // get a FileDescriptor number from the OS to use for our socket + result.server_socket = socket(PF_INET, SOCK_DGRAM, 0); + if (result.server_socket < 0) { + return result; + } + // to let us immediately kill and restart server + i32 optval = 1; + setsockopt(result.server_socket, SOL_SOCKET, SO_REUSEADDR, (const void *)&optval, sizeof(i32)); + + result.ready = bind(result.server_socket, (struct sockaddr *)&result.server_address, sizeof(result.server_address)) >= 0; + + return result; +} + +UDPClient createUDPClient(u16 server_port, str addr) { + UDPClient result = {0}; + // define the address we'll be listening on + result.server_address.sin_family = AF_INET; + if (addr == 0) { + result.server_address.sin_addr.s_addr = inet_addr("127.0.0.1"); + } else { + result.server_address.sin_addr.s_addr = inet_addr(addr); + } + result.server_address.sin_port = htons(server_port); + + // get a FileDescriptor number from the OS to use for our socket + result.socket = socket(PF_INET, SOCK_DGRAM, 0); + if (result.socket < 0) { + return result; + } + + struct sockaddr_in client_address = {0}; + client_address.sin_family = AF_INET; + client_address.sin_addr.s_addr = htonl(INADDR_ANY); + client_address.sin_port = 0; + result.ready = bind(result.socket, (struct sockaddr *)&client_address, sizeof(client_address)) >= 0; + struct sockaddr_in empty_addr; + socklen_t addr_len = sizeof(empty_addr); + getsockname(result.socket, (struct sockaddr *)&empty_addr, &addr_len); + result.client_port = ntohs(empty_addr.sin_port); + return result; +} + +void infiniteReadUDPServer(UDPServer* server, void (*handleMessage)(u8* udp_message, u32 udp_len, SocketAddress sending_address, i32 socket)) { + u8 message_buffer[UDP_MAX_MESSAGE_LEN] = {0}; + i32 bytes_recieved = 0; + SocketAddress client_address = {0}; + i32 addrlen = sizeof(struct sockaddr); + while (true) { + bytes_recieved = recvfrom(server->server_socket, message_buffer, UDP_MAX_MESSAGE_LEN, 0, (struct sockaddr *)&client_address, (socklen_t*)&addrlen); + + //gethostbyaddr: determine who sent the datagram + //struct hostent* hostp = gethostbyaddr( + // (const char *)&client_address.sin_addr.s_addr, + // sizeof(client_address.sin_addr.s_addr), + // AF_INET + //); + + //ptr printable_host_IP_address_string = inet_ntoa(client_address.sin_addr); + + handleMessage(message_buffer, bytes_recieved, client_address, server->server_socket); + MemoryZero(message_buffer, UDP_MAX_MESSAGE_LEN); + } +} + +// TODO: sendall() to handle cases when the sendto() bytes return value is less than the intended bytes to send... stupid kernel fuckin wit us. +i32 sendUDPu8List(i32 using_socket, SocketAddress* to, u8List* message) { + return sendto( + using_socket, + message->items, + message->length, + 0, + (const struct sockaddr *)to, + sizeof(struct sockaddr) + ); +} + +i32 sendUDPMessage(UDPServer* to, u8* message, u32 len) { + return sendto(to->server_socket, message, len, 0, (struct sockaddr *)&to->server_address, sizeof(struct sockaddr)); +} diff --git a/src/lib/stb_image.h b/src/lib/stb_image.h new file mode 100644 index 0000000..d4b5c87 --- /dev/null +++ b/src/lib/stb_image.h @@ -0,0 +1,7988 @@ +/* stb_image - v2.30 - public domain image loader - http://nothings.org/stb + no warranty implied; use at your own risk + + Do this: + #define STB_IMAGE_IMPLEMENTATION + before you include this file in *one* C or C++ file to create the implementation. + + // i.e. it should look like this: + #include ... + #include ... + #include ... + #define STB_IMAGE_IMPLEMENTATION + #include "stb_image.h" + + You can #define STBI_ASSERT(x) before the #include to avoid using assert.h. + And #define STBI_MALLOC, STBI_REALLOC, and STBI_FREE to avoid using malloc,realloc,free + + + QUICK NOTES: + Primarily of interest to game developers and other people who can + avoid problematic images and only need the trivial interface + + JPEG baseline & progressive (12 bpc/arithmetic not supported, same as stock IJG lib) + PNG 1/2/4/8/16-bit-per-channel + + TGA (not sure what subset, if a subset) + BMP non-1bpp, non-RLE + PSD (composited view only, no extra channels, 8/16 bit-per-channel) + + GIF (*comp always reports as 4-channel) + HDR (radiance rgbE format) + PIC (Softimage PIC) + PNM (PPM and PGM binary only) + + Animated GIF still needs a proper API, but here's one way to do it: + http://gist.github.com/urraka/685d9a6340b26b830d49 + + - decode from memory or through FILE (define STBI_NO_STDIO to remove code) + - decode from arbitrary I/O callbacks + - SIMD acceleration on x86/x64 (SSE2) and ARM (NEON) + + Full documentation under "DOCUMENTATION" below. + + +LICENSE + + See end of file for license information. + +RECENT REVISION HISTORY: + + 2.30 (2024-05-31) avoid erroneous gcc warning + 2.29 (2023-05-xx) optimizations + 2.28 (2023-01-29) many error fixes, security errors, just tons of stuff + 2.27 (2021-07-11) document stbi_info better, 16-bit PNM support, bug fixes + 2.26 (2020-07-13) many minor fixes + 2.25 (2020-02-02) fix warnings + 2.24 (2020-02-02) fix warnings; thread-local failure_reason and flip_vertically + 2.23 (2019-08-11) fix clang static analysis warning + 2.22 (2019-03-04) gif fixes, fix warnings + 2.21 (2019-02-25) fix typo in comment + 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs + 2.19 (2018-02-11) fix warning + 2.18 (2018-01-30) fix warnings + 2.17 (2018-01-29) bugfix, 1-bit BMP, 16-bitness query, fix warnings + 2.16 (2017-07-23) all functions have 16-bit variants; optimizations; bugfixes + 2.15 (2017-03-18) fix png-1,2,4; all Imagenet JPGs; no runtime SSE detection on GCC + 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs + 2.13 (2016-12-04) experimental 16-bit API, only for PNG so far; fixes + 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes + 2.11 (2016-04-02) 16-bit PNGS; enable SSE2 in non-gcc x64 + RGB-format JPEG; remove white matting in PSD; + allocate large structures on the stack; + correct channel count for PNG & BMP + 2.10 (2016-01-22) avoid warning introduced in 2.09 + 2.09 (2016-01-16) 16-bit TGA; comments in PNM files; STBI_REALLOC_SIZED + + See end of file for full revision history. + + + ============================ Contributors ========================= + + Image formats Extensions, features + Sean Barrett (jpeg, png, bmp) Jetro Lauha (stbi_info) + Nicolas Schulz (hdr, psd) Martin "SpartanJ" Golini (stbi_info) + Jonathan Dummer (tga) James "moose2000" Brown (iPhone PNG) + Jean-Marc Lienher (gif) Ben "Disch" Wenger (io callbacks) + Tom Seddon (pic) Omar Cornut (1/2/4-bit PNG) + Thatcher Ulrich (psd) Nicolas Guillemot (vertical flip) + Ken Miller (pgm, ppm) Richard Mitton (16-bit PSD) + github:urraka (animated gif) Junggon Kim (PNM comments) + Christopher Forseth (animated gif) Daniel Gibson (16-bit TGA) + socks-the-fox (16-bit PNG) + Jeremy Sawicki (handle all ImageNet JPGs) + Optimizations & bugfixes Mikhail Morozov (1-bit BMP) + Fabian "ryg" Giesen Anael Seghezzi (is-16-bit query) + Arseny Kapoulkine Simon Breuss (16-bit PNM) + John-Mark Allen + Carmelo J Fdez-Aguera + + Bug & warning fixes + Marc LeBlanc David Woo Guillaume George Martins Mozeiko + Christpher Lloyd Jerry Jansson Joseph Thomson Blazej Dariusz Roszkowski + Phil Jordan Dave Moore Roy Eltham + Hayaki Saito Nathan Reed Won Chun + Luke Graham Johan Duparc Nick Verigakis the Horde3D community + Thomas Ruf Ronny Chevalier github:rlyeh + Janez Zemva John Bartholomew Michal Cichon github:romigrou + Jonathan Blow Ken Hamada Tero Hanninen github:svdijk + Eugene Golushkov Laurent Gomila Cort Stratton github:snagar + Aruelien Pocheville Sergio Gonzalez Thibault Reuille github:Zelex + Cass Everitt Ryamond Barbiero github:grim210 + Paul Du Bois Engin Manap Aldo Culquicondor github:sammyhw + Philipp Wiesemann Dale Weiler Oriol Ferrer Mesia github:phprus + Josh Tobin Neil Bickford Matthew Gregan github:poppolopoppo + Julian Raschke Gregory Mullen Christian Floisand github:darealshinji + Baldur Karlsson Kevin Schmidt JR Smith github:Michaelangel007 + Brad Weinberger Matvey Cherevko github:mosra + Luca Sas Alexander Veselov Zack Middleton [reserved] + Ryan C. Gordon [reserved] [reserved] + DO NOT ADD YOUR NAME HERE + + Jacko Dirks + + To add your name to the credits, pick a random blank space in the middle and fill it. + 80% of merge conflicts on stb PRs are due to people adding their name at the end + of the credits. +*/ + +#ifndef STBI_INCLUDE_STB_IMAGE_H +#define STBI_INCLUDE_STB_IMAGE_H + +// DOCUMENTATION +// +// Limitations: +// - no 12-bit-per-channel JPEG +// - no JPEGs with arithmetic coding +// - GIF always returns *comp=4 +// +// Basic usage (see HDR discussion below for HDR usage): +// int x,y,n; +// unsigned char *data = stbi_load(filename, &x, &y, &n, 0); +// // ... process data if not NULL ... +// // ... x = width, y = height, n = # 8-bit components per pixel ... +// // ... replace '0' with '1'..'4' to force that many components per pixel +// // ... but 'n' will always be the number that it would have been if you said 0 +// stbi_image_free(data); +// +// Standard parameters: +// int *x -- outputs image width in pixels +// int *y -- outputs image height in pixels +// int *channels_in_file -- outputs # of image components in image file +// int desired_channels -- if non-zero, # of image components requested in result +// +// The return value from an image loader is an 'unsigned char *' which points +// to the pixel data, or NULL on an allocation failure or if the image is +// corrupt or invalid. The pixel data consists of *y scanlines of *x pixels, +// with each pixel consisting of N interleaved 8-bit components; the first +// pixel pointed to is top-left-most in the image. There is no padding between +// image scanlines or between pixels, regardless of format. The number of +// components N is 'desired_channels' if desired_channels is non-zero, or +// *channels_in_file otherwise. If desired_channels is non-zero, +// *channels_in_file has the number of components that _would_ have been +// output otherwise. E.g. if you set desired_channels to 4, you will always +// get RGBA output, but you can check *channels_in_file to see if it's trivially +// opaque because e.g. there were only 3 channels in the source image. +// +// An output image with N components has the following components interleaved +// in this order in each pixel: +// +// N=#comp components +// 1 grey +// 2 grey, alpha +// 3 red, green, blue +// 4 red, green, blue, alpha +// +// If image loading fails for any reason, the return value will be NULL, +// and *x, *y, *channels_in_file will be unchanged. The function +// stbi_failure_reason() can be queried for an extremely brief, end-user +// unfriendly explanation of why the load failed. Define STBI_NO_FAILURE_STRINGS +// to avoid compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly +// more user-friendly ones. +// +// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized. +// +// To query the width, height and component count of an image without having to +// decode the full file, you can use the stbi_info family of functions: +// +// int x,y,n,ok; +// ok = stbi_info(filename, &x, &y, &n); +// // returns ok=1 and sets x, y, n if image is a supported format, +// // 0 otherwise. +// +// Note that stb_image pervasively uses ints in its public API for sizes, +// including sizes of memory buffers. This is now part of the API and thus +// hard to change without causing breakage. As a result, the various image +// loaders all have certain limits on image size; these differ somewhat +// by format but generally boil down to either just under 2GB or just under +// 1GB. When the decoded image would be larger than this, stb_image decoding +// will fail. +// +// Additionally, stb_image will reject image files that have any of their +// dimensions set to a larger value than the configurable STBI_MAX_DIMENSIONS, +// which defaults to 2**24 = 16777216 pixels. Due to the above memory limit, +// the only way to have an image with such dimensions load correctly +// is for it to have a rather extreme aspect ratio. Either way, the +// assumption here is that such larger images are likely to be malformed +// or malicious. If you do need to load an image with individual dimensions +// larger than that, and it still fits in the overall size limit, you can +// #define STBI_MAX_DIMENSIONS on your own to be something larger. +// +// =========================================================================== +// +// UNICODE: +// +// If compiling for Windows and you wish to use Unicode filenames, compile +// with +// #define STBI_WINDOWS_UTF8 +// and pass utf8-encoded filenames. Call stbi_convert_wchar_to_utf8 to convert +// Windows wchar_t filenames to utf8. +// +// =========================================================================== +// +// Philosophy +// +// stb libraries are designed with the following priorities: +// +// 1. easy to use +// 2. easy to maintain +// 3. good performance +// +// Sometimes I let "good performance" creep up in priority over "easy to maintain", +// and for best performance I may provide less-easy-to-use APIs that give higher +// performance, in addition to the easy-to-use ones. Nevertheless, it's important +// to keep in mind that from the standpoint of you, a client of this library, +// all you care about is #1 and #3, and stb libraries DO NOT emphasize #3 above all. +// +// Some secondary priorities arise directly from the first two, some of which +// provide more explicit reasons why performance can't be emphasized. +// +// - Portable ("ease of use") +// - Small source code footprint ("easy to maintain") +// - No dependencies ("ease of use") +// +// =========================================================================== +// +// I/O callbacks +// +// I/O callbacks allow you to read from arbitrary sources, like packaged +// files or some other source. Data read from callbacks are processed +// through a small internal buffer (currently 128 bytes) to try to reduce +// overhead. +// +// The three functions you must define are "read" (reads some bytes of data), +// "skip" (skips some bytes of data), "eof" (reports if the stream is at the end). +// +// =========================================================================== +// +// SIMD support +// +// The JPEG decoder will try to automatically use SIMD kernels on x86 when +// supported by the compiler. For ARM Neon support, you must explicitly +// request it. +// +// (The old do-it-yourself SIMD API is no longer supported in the current +// code.) +// +// On x86, SSE2 will automatically be used when available based on a run-time +// test; if not, the generic C versions are used as a fall-back. On ARM targets, +// the typical path is to have separate builds for NEON and non-NEON devices +// (at least this is true for iOS and Android). Therefore, the NEON support is +// toggled by a build flag: define STBI_NEON to get NEON loops. +// +// If for some reason you do not want to use any of SIMD code, or if +// you have issues compiling it, you can disable it entirely by +// defining STBI_NO_SIMD. +// +// =========================================================================== +// +// HDR image support (disable by defining STBI_NO_HDR) +// +// stb_image supports loading HDR images in general, and currently the Radiance +// .HDR file format specifically. You can still load any file through the existing +// interface; if you attempt to load an HDR file, it will be automatically remapped +// to LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1; +// both of these constants can be reconfigured through this interface: +// +// stbi_hdr_to_ldr_gamma(2.2f); +// stbi_hdr_to_ldr_scale(1.0f); +// +// (note, do not use _inverse_ constants; stbi_image will invert them +// appropriately). +// +// Additionally, there is a new, parallel interface for loading files as +// (linear) floats to preserve the full dynamic range: +// +// float *data = stbi_loadf(filename, &x, &y, &n, 0); +// +// If you load LDR images through this interface, those images will +// be promoted to floating point values, run through the inverse of +// constants corresponding to the above: +// +// stbi_ldr_to_hdr_scale(1.0f); +// stbi_ldr_to_hdr_gamma(2.2f); +// +// Finally, given a filename (or an open file or memory block--see header +// file for details) containing image data, you can query for the "most +// appropriate" interface to use (that is, whether the image is HDR or +// not), using: +// +// stbi_is_hdr(char *filename); +// +// =========================================================================== +// +// iPhone PNG support: +// +// We optionally support converting iPhone-formatted PNGs (which store +// premultiplied BGRA) back to RGB, even though they're internally encoded +// differently. To enable this conversion, call +// stbi_convert_iphone_png_to_rgb(1). +// +// Call stbi_set_unpremultiply_on_load(1) as well to force a divide per +// pixel to remove any premultiplied alpha *only* if the image file explicitly +// says there's premultiplied data (currently only happens in iPhone images, +// and only if iPhone convert-to-rgb processing is on). +// +// =========================================================================== +// +// ADDITIONAL CONFIGURATION +// +// - You can suppress implementation of any of the decoders to reduce +// your code footprint by #defining one or more of the following +// symbols before creating the implementation. +// +// STBI_NO_JPEG +// STBI_NO_PNG +// STBI_NO_BMP +// STBI_NO_PSD +// STBI_NO_TGA +// STBI_NO_GIF +// STBI_NO_HDR +// STBI_NO_PIC +// STBI_NO_PNM (.ppm and .pgm) +// +// - You can request *only* certain decoders and suppress all other ones +// (this will be more forward-compatible, as addition of new decoders +// doesn't require you to disable them explicitly): +// +// STBI_ONLY_JPEG +// STBI_ONLY_PNG +// STBI_ONLY_BMP +// STBI_ONLY_PSD +// STBI_ONLY_TGA +// STBI_ONLY_GIF +// STBI_ONLY_HDR +// STBI_ONLY_PIC +// STBI_ONLY_PNM (.ppm and .pgm) +// +// - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still +// want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB +// +// - If you define STBI_MAX_DIMENSIONS, stb_image will reject images greater +// than that size (in either width or height) without further processing. +// This is to let programs in the wild set an upper bound to prevent +// denial-of-service attacks on untrusted data, as one could generate a +// valid image of gigantic dimensions and force stb_image to allocate a +// huge block of memory and spend disproportionate time decoding it. By +// default this is set to (1 << 24), which is 16777216, but that's still +// very big. + +#ifndef STBI_NO_STDIO +#include +#endif // STBI_NO_STDIO + +#define STBI_VERSION 1 + +enum +{ + STBI_default = 0, // only used for desired_channels + + STBI_grey = 1, + STBI_grey_alpha = 2, + STBI_rgb = 3, + STBI_rgb_alpha = 4 +}; + +#include +typedef unsigned char stbi_uc; +typedef unsigned short stbi_us; + +#ifdef __cplusplus +extern "C" { +#endif + +#ifndef STBIDEF +#ifdef STB_IMAGE_STATIC +#define STBIDEF static +#else +#define STBIDEF extern +#endif +#endif + +////////////////////////////////////////////////////////////////////////////// +// +// PRIMARY API - works on images of any type +// + +// +// load image by filename, open file, or memory buffer +// + +typedef struct +{ + int (*read) (void *user,char *data,int size); // fill 'data' with 'size' bytes. return number of bytes actually read + void (*skip) (void *user,int n); // skip the next 'n' bytes, or 'unget' the last -n bytes if negative + int (*eof) (void *user); // returns nonzero if we are at end of file/data +} stbi_io_callbacks; + +//////////////////////////////////// +// +// 8-bits-per-channel interface +// + +STBIDEF stbi_uc *stbi_load_from_memory (stbi_uc const *buffer, int len , int *x, int *y, int *channels_in_file, int desired_channels); +STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk , void *user, int *x, int *y, int *channels_in_file, int desired_channels); + +#ifndef STBI_NO_STDIO +STBIDEF stbi_uc *stbi_load (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels); +STBIDEF stbi_uc *stbi_load_from_file (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels); +// for stbi_load_from_file, file pointer is left pointing immediately after image +#endif + +#ifndef STBI_NO_GIF +STBIDEF stbi_uc *stbi_load_gif_from_memory(stbi_uc const *buffer, int len, int **delays, int *x, int *y, int *z, int *comp, int req_comp); +#endif + +#ifdef STBI_WINDOWS_UTF8 +STBIDEF int stbi_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input); +#endif + +//////////////////////////////////// +// +// 16-bits-per-channel interface +// + +STBIDEF stbi_us *stbi_load_16_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels); +STBIDEF stbi_us *stbi_load_16_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels); + +#ifndef STBI_NO_STDIO +STBIDEF stbi_us *stbi_load_16 (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels); +STBIDEF stbi_us *stbi_load_from_file_16(FILE *f, int *x, int *y, int *channels_in_file, int desired_channels); +#endif + +//////////////////////////////////// +// +// float-per-channel interface +// +#ifndef STBI_NO_LINEAR + STBIDEF float *stbi_loadf_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels); + STBIDEF float *stbi_loadf_from_callbacks (stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels); + + #ifndef STBI_NO_STDIO + STBIDEF float *stbi_loadf (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels); + STBIDEF float *stbi_loadf_from_file (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels); + #endif +#endif + +#ifndef STBI_NO_HDR + STBIDEF void stbi_hdr_to_ldr_gamma(float gamma); + STBIDEF void stbi_hdr_to_ldr_scale(float scale); +#endif // STBI_NO_HDR + +#ifndef STBI_NO_LINEAR + STBIDEF void stbi_ldr_to_hdr_gamma(float gamma); + STBIDEF void stbi_ldr_to_hdr_scale(float scale); +#endif // STBI_NO_LINEAR + +// stbi_is_hdr is always defined, but always returns false if STBI_NO_HDR +STBIDEF int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user); +STBIDEF int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len); +#ifndef STBI_NO_STDIO +STBIDEF int stbi_is_hdr (char const *filename); +STBIDEF int stbi_is_hdr_from_file(FILE *f); +#endif // STBI_NO_STDIO + + +// get a VERY brief reason for failure +// on most compilers (and ALL modern mainstream compilers) this is threadsafe +STBIDEF const char *stbi_failure_reason (void); + +// free the loaded image -- this is just free() +STBIDEF void stbi_image_free (void *retval_from_stbi_load); + +// get image dimensions & components without fully decoding +STBIDEF int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp); +STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp); +STBIDEF int stbi_is_16_bit_from_memory(stbi_uc const *buffer, int len); +STBIDEF int stbi_is_16_bit_from_callbacks(stbi_io_callbacks const *clbk, void *user); + +#ifndef STBI_NO_STDIO +STBIDEF int stbi_info (char const *filename, int *x, int *y, int *comp); +STBIDEF int stbi_info_from_file (FILE *f, int *x, int *y, int *comp); +STBIDEF int stbi_is_16_bit (char const *filename); +STBIDEF int stbi_is_16_bit_from_file(FILE *f); +#endif + + + +// for image formats that explicitly notate that they have premultiplied alpha, +// we just return the colors as stored in the file. set this flag to force +// unpremultiplication. results are undefined if the unpremultiply overflow. +STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply); + +// indicate whether we should process iphone images back to canonical format, +// or just pass them through "as-is" +STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert); + +// flip the image vertically, so the first pixel in the output array is the bottom left +STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip); + +// as above, but only applies to images loaded on the thread that calls the function +// this function is only available if your compiler supports thread-local variables; +// calling it will fail to link if your compiler doesn't +STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply); +STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert); +STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip); + +// ZLIB client - used by PNG, available for other purposes + +STBIDEF char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen); +STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, int len, int initial_size, int *outlen, int parse_header); +STBIDEF char *stbi_zlib_decode_malloc(const char *buffer, int len, int *outlen); +STBIDEF int stbi_zlib_decode_buffer(char *obuffer, int olen, const char *ibuffer, int ilen); + +STBIDEF char *stbi_zlib_decode_noheader_malloc(const char *buffer, int len, int *outlen); +STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen); + + +#ifdef __cplusplus +} +#endif + +// +// +//// end header file ///////////////////////////////////////////////////// +#endif // STBI_INCLUDE_STB_IMAGE_H + +#ifdef STB_IMAGE_IMPLEMENTATION + +#if defined(STBI_ONLY_JPEG) || defined(STBI_ONLY_PNG) || defined(STBI_ONLY_BMP) \ + || defined(STBI_ONLY_TGA) || defined(STBI_ONLY_GIF) || defined(STBI_ONLY_PSD) \ + || defined(STBI_ONLY_HDR) || defined(STBI_ONLY_PIC) || defined(STBI_ONLY_PNM) \ + || defined(STBI_ONLY_ZLIB) + #ifndef STBI_ONLY_JPEG + #define STBI_NO_JPEG + #endif + #ifndef STBI_ONLY_PNG + #define STBI_NO_PNG + #endif + #ifndef STBI_ONLY_BMP + #define STBI_NO_BMP + #endif + #ifndef STBI_ONLY_PSD + #define STBI_NO_PSD + #endif + #ifndef STBI_ONLY_TGA + #define STBI_NO_TGA + #endif + #ifndef STBI_ONLY_GIF + #define STBI_NO_GIF + #endif + #ifndef STBI_ONLY_HDR + #define STBI_NO_HDR + #endif + #ifndef STBI_ONLY_PIC + #define STBI_NO_PIC + #endif + #ifndef STBI_ONLY_PNM + #define STBI_NO_PNM + #endif +#endif + +#if defined(STBI_NO_PNG) && !defined(STBI_SUPPORT_ZLIB) && !defined(STBI_NO_ZLIB) +#define STBI_NO_ZLIB +#endif + + +#include +#include // ptrdiff_t on osx +#include +#include +#include + +#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) +#include // ldexp, pow +#endif + +#ifndef STBI_NO_STDIO +#include +#endif + +#ifndef STBI_ASSERT +#include +#define STBI_ASSERT(x) assert(x) +#endif + +#ifdef __cplusplus +#define STBI_EXTERN extern "C" +#else +#define STBI_EXTERN extern +#endif + + +#ifndef _MSC_VER + #ifdef __cplusplus + #define stbi_inline inline + #else + #define stbi_inline + #endif +#else + #define stbi_inline __forceinline +#endif + +#ifndef STBI_NO_THREAD_LOCALS + #if defined(__cplusplus) && __cplusplus >= 201103L + #define STBI_THREAD_LOCAL thread_local + #elif defined(__GNUC__) && __GNUC__ < 5 + #define STBI_THREAD_LOCAL __thread + #elif defined(_MSC_VER) + #define STBI_THREAD_LOCAL __declspec(thread) + #elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 201112L && !defined(__STDC_NO_THREADS__) + #define STBI_THREAD_LOCAL _Thread_local + #endif + + #ifndef STBI_THREAD_LOCAL + #if defined(__GNUC__) + #define STBI_THREAD_LOCAL __thread + #endif + #endif +#endif + +#if defined(_MSC_VER) || defined(__SYMBIAN32__) +typedef unsigned short stbi__uint16; +typedef signed short stbi__int16; +typedef unsigned int stbi__uint32; +typedef signed int stbi__int32; +#else +#include +typedef uint16_t stbi__uint16; +typedef int16_t stbi__int16; +typedef uint32_t stbi__uint32; +typedef int32_t stbi__int32; +#endif + +// should produce compiler error if size is wrong +typedef unsigned char validate_uint32[sizeof(stbi__uint32)==4 ? 1 : -1]; + +#ifdef _MSC_VER +#define STBI_NOTUSED(v) (void)(v) +#else +#define STBI_NOTUSED(v) (void)sizeof(v) +#endif + +#ifdef _MSC_VER +#define STBI_HAS_LROTL +#endif + +#ifdef STBI_HAS_LROTL + #define stbi_lrot(x,y) _lrotl(x,y) +#else + #define stbi_lrot(x,y) (((x) << (y)) | ((x) >> (-(y) & 31))) +#endif + +#if defined(STBI_MALLOC) && defined(STBI_FREE) && (defined(STBI_REALLOC) || defined(STBI_REALLOC_SIZED)) +// ok +#elif !defined(STBI_MALLOC) && !defined(STBI_FREE) && !defined(STBI_REALLOC) && !defined(STBI_REALLOC_SIZED) +// ok +#else +#error "Must define all or none of STBI_MALLOC, STBI_FREE, and STBI_REALLOC (or STBI_REALLOC_SIZED)." +#endif + +#ifndef STBI_MALLOC +#define STBI_MALLOC(sz) malloc(sz) +#define STBI_REALLOC(p,newsz) realloc(p,newsz) +#define STBI_FREE(p) free(p) +#endif + +#ifndef STBI_REALLOC_SIZED +#define STBI_REALLOC_SIZED(p,oldsz,newsz) STBI_REALLOC(p,newsz) +#endif + +// x86/x64 detection +#if defined(__x86_64__) || defined(_M_X64) +#define STBI__X64_TARGET +#elif defined(__i386) || defined(_M_IX86) +#define STBI__X86_TARGET +#endif + +#if defined(__GNUC__) && defined(STBI__X86_TARGET) && !defined(__SSE2__) && !defined(STBI_NO_SIMD) +// gcc doesn't support sse2 intrinsics unless you compile with -msse2, +// which in turn means it gets to use SSE2 everywhere. This is unfortunate, +// but previous attempts to provide the SSE2 functions with runtime +// detection caused numerous issues. The way architecture extensions are +// exposed in GCC/Clang is, sadly, not really suited for one-file libs. +// New behavior: if compiled with -msse2, we use SSE2 without any +// detection; if not, we don't use it at all. +#define STBI_NO_SIMD +#endif + +#if defined(__MINGW32__) && defined(STBI__X86_TARGET) && !defined(STBI_MINGW_ENABLE_SSE2) && !defined(STBI_NO_SIMD) +// Note that __MINGW32__ doesn't actually mean 32-bit, so we have to avoid STBI__X64_TARGET +// +// 32-bit MinGW wants ESP to be 16-byte aligned, but this is not in the +// Windows ABI and VC++ as well as Windows DLLs don't maintain that invariant. +// As a result, enabling SSE2 on 32-bit MinGW is dangerous when not +// simultaneously enabling "-mstackrealign". +// +// See https://github.com/nothings/stb/issues/81 for more information. +// +// So default to no SSE2 on 32-bit MinGW. If you've read this far and added +// -mstackrealign to your build settings, feel free to #define STBI_MINGW_ENABLE_SSE2. +#define STBI_NO_SIMD +#endif + +#if !defined(STBI_NO_SIMD) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET)) +#define STBI_SSE2 +#include + +#ifdef _MSC_VER + +#if _MSC_VER >= 1400 // not VC6 +#include // __cpuid +static int stbi__cpuid3(void) +{ + int info[4]; + __cpuid(info,1); + return info[3]; +} +#else +static int stbi__cpuid3(void) +{ + int res; + __asm { + mov eax,1 + cpuid + mov res,edx + } + return res; +} +#endif + +#define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name + +#if !defined(STBI_NO_JPEG) && defined(STBI_SSE2) +static int stbi__sse2_available(void) +{ + int info3 = stbi__cpuid3(); + return ((info3 >> 26) & 1) != 0; +} +#endif + +#else // assume GCC-style if not VC++ +#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16))) + +#if !defined(STBI_NO_JPEG) && defined(STBI_SSE2) +static int stbi__sse2_available(void) +{ + // If we're even attempting to compile this on GCC/Clang, that means + // -msse2 is on, which means the compiler is allowed to use SSE2 + // instructions at will, and so are we. + return 1; +} +#endif + +#endif +#endif + +// ARM NEON +#if defined(STBI_NO_SIMD) && defined(STBI_NEON) +#undef STBI_NEON +#endif + +#ifdef STBI_NEON +#include +#ifdef _MSC_VER +#define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name +#else +#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16))) +#endif +#endif + +#ifndef STBI_SIMD_ALIGN +#define STBI_SIMD_ALIGN(type, name) type name +#endif + +#ifndef STBI_MAX_DIMENSIONS +#define STBI_MAX_DIMENSIONS (1 << 24) +#endif + +/////////////////////////////////////////////// +// +// stbi__context struct and start_xxx functions + +// stbi__context structure is our basic context used by all images, so it +// contains all the IO context, plus some basic image information +typedef struct +{ + stbi__uint32 img_x, img_y; + int img_n, img_out_n; + + stbi_io_callbacks io; + void *io_user_data; + + int read_from_callbacks; + int buflen; + stbi_uc buffer_start[128]; + int callback_already_read; + + stbi_uc *img_buffer, *img_buffer_end; + stbi_uc *img_buffer_original, *img_buffer_original_end; +} stbi__context; + + +static void stbi__refill_buffer(stbi__context *s); + +// initialize a memory-decode context +static void stbi__start_mem(stbi__context *s, stbi_uc const *buffer, int len) +{ + s->io.read = NULL; + s->read_from_callbacks = 0; + s->callback_already_read = 0; + s->img_buffer = s->img_buffer_original = (stbi_uc *) buffer; + s->img_buffer_end = s->img_buffer_original_end = (stbi_uc *) buffer+len; +} + +// initialize a callback-based context +static void stbi__start_callbacks(stbi__context *s, stbi_io_callbacks *c, void *user) +{ + s->io = *c; + s->io_user_data = user; + s->buflen = sizeof(s->buffer_start); + s->read_from_callbacks = 1; + s->callback_already_read = 0; + s->img_buffer = s->img_buffer_original = s->buffer_start; + stbi__refill_buffer(s); + s->img_buffer_original_end = s->img_buffer_end; +} + +#ifndef STBI_NO_STDIO + +static int stbi__stdio_read(void *user, char *data, int size) +{ + return (int) fread(data,1,size,(FILE*) user); +} + +static void stbi__stdio_skip(void *user, int n) +{ + int ch; + fseek((FILE*) user, n, SEEK_CUR); + ch = fgetc((FILE*) user); /* have to read a byte to reset feof()'s flag */ + if (ch != EOF) { + ungetc(ch, (FILE *) user); /* push byte back onto stream if valid. */ + } +} + +static int stbi__stdio_eof(void *user) +{ + return feof((FILE*) user) || ferror((FILE *) user); +} + +static stbi_io_callbacks stbi__stdio_callbacks = +{ + stbi__stdio_read, + stbi__stdio_skip, + stbi__stdio_eof, +}; + +static void stbi__start_file(stbi__context *s, FILE *f) +{ + stbi__start_callbacks(s, &stbi__stdio_callbacks, (void *) f); +} + +//static void stop_file(stbi__context *s) { } + +#endif // !STBI_NO_STDIO + +static void stbi__rewind(stbi__context *s) +{ + // conceptually rewind SHOULD rewind to the beginning of the stream, + // but we just rewind to the beginning of the initial buffer, because + // we only use it after doing 'test', which only ever looks at at most 92 bytes + s->img_buffer = s->img_buffer_original; + s->img_buffer_end = s->img_buffer_original_end; +} + +enum +{ + STBI_ORDER_RGB, + STBI_ORDER_BGR +}; + +typedef struct +{ + int bits_per_channel; + int num_channels; + int channel_order; +} stbi__result_info; + +#ifndef STBI_NO_JPEG +static int stbi__jpeg_test(stbi__context *s); +static void *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_PNG +static int stbi__png_test(stbi__context *s); +static void *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp); +static int stbi__png_is16(stbi__context *s); +#endif + +#ifndef STBI_NO_BMP +static int stbi__bmp_test(stbi__context *s); +static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_TGA +static int stbi__tga_test(stbi__context *s); +static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_PSD +static int stbi__psd_test(stbi__context *s); +static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc); +static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp); +static int stbi__psd_is16(stbi__context *s); +#endif + +#ifndef STBI_NO_HDR +static int stbi__hdr_test(stbi__context *s); +static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_PIC +static int stbi__pic_test(stbi__context *s); +static void *stbi__pic_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_GIF +static int stbi__gif_test(stbi__context *s); +static void *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static void *stbi__load_gif_main(stbi__context *s, int **delays, int *x, int *y, int *z, int *comp, int req_comp); +static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_PNM +static int stbi__pnm_test(stbi__context *s); +static void *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri); +static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp); +static int stbi__pnm_is16(stbi__context *s); +#endif + +static +#ifdef STBI_THREAD_LOCAL +STBI_THREAD_LOCAL +#endif +const char *stbi__g_failure_reason; + +STBIDEF const char *stbi_failure_reason(void) +{ + return stbi__g_failure_reason; +} + +#ifndef STBI_NO_FAILURE_STRINGS +static int stbi__err(const char *str) +{ + stbi__g_failure_reason = str; + return 0; +} +#endif + +static void *stbi__malloc(size_t size) +{ + return STBI_MALLOC(size); +} + +// stb_image uses ints pervasively, including for offset calculations. +// therefore the largest decoded image size we can support with the +// current code, even on 64-bit targets, is INT_MAX. this is not a +// significant limitation for the intended use case. +// +// we do, however, need to make sure our size calculations don't +// overflow. hence a few helper functions for size calculations that +// multiply integers together, making sure that they're non-negative +// and no overflow occurs. + +// return 1 if the sum is valid, 0 on overflow. +// negative terms are considered invalid. +static int stbi__addsizes_valid(int a, int b) +{ + if (b < 0) return 0; + // now 0 <= b <= INT_MAX, hence also + // 0 <= INT_MAX - b <= INTMAX. + // And "a + b <= INT_MAX" (which might overflow) is the + // same as a <= INT_MAX - b (no overflow) + return a <= INT_MAX - b; +} + +// returns 1 if the product is valid, 0 on overflow. +// negative factors are considered invalid. +static int stbi__mul2sizes_valid(int a, int b) +{ + if (a < 0 || b < 0) return 0; + if (b == 0) return 1; // mul-by-0 is always safe + // portable way to check for no overflows in a*b + return a <= INT_MAX/b; +} + +#if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR) +// returns 1 if "a*b + add" has no negative terms/factors and doesn't overflow +static int stbi__mad2sizes_valid(int a, int b, int add) +{ + return stbi__mul2sizes_valid(a, b) && stbi__addsizes_valid(a*b, add); +} +#endif + +// returns 1 if "a*b*c + add" has no negative terms/factors and doesn't overflow +static int stbi__mad3sizes_valid(int a, int b, int c, int add) +{ + return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a*b, c) && + stbi__addsizes_valid(a*b*c, add); +} + +// returns 1 if "a*b*c*d + add" has no negative terms/factors and doesn't overflow +#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM) +static int stbi__mad4sizes_valid(int a, int b, int c, int d, int add) +{ + return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a*b, c) && + stbi__mul2sizes_valid(a*b*c, d) && stbi__addsizes_valid(a*b*c*d, add); +} +#endif + +#if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR) +// mallocs with size overflow checking +static void *stbi__malloc_mad2(int a, int b, int add) +{ + if (!stbi__mad2sizes_valid(a, b, add)) return NULL; + return stbi__malloc(a*b + add); +} +#endif + +static void *stbi__malloc_mad3(int a, int b, int c, int add) +{ + if (!stbi__mad3sizes_valid(a, b, c, add)) return NULL; + return stbi__malloc(a*b*c + add); +} + +#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM) +static void *stbi__malloc_mad4(int a, int b, int c, int d, int add) +{ + if (!stbi__mad4sizes_valid(a, b, c, d, add)) return NULL; + return stbi__malloc(a*b*c*d + add); +} +#endif + +// returns 1 if the sum of two signed ints is valid (between -2^31 and 2^31-1 inclusive), 0 on overflow. +static int stbi__addints_valid(int a, int b) +{ + if ((a >= 0) != (b >= 0)) return 1; // a and b have different signs, so no overflow + if (a < 0 && b < 0) return a >= INT_MIN - b; // same as a + b >= INT_MIN; INT_MIN - b cannot overflow since b < 0. + return a <= INT_MAX - b; +} + +// returns 1 if the product of two ints fits in a signed short, 0 on overflow. +static int stbi__mul2shorts_valid(int a, int b) +{ + if (b == 0 || b == -1) return 1; // multiplication by 0 is always 0; check for -1 so SHRT_MIN/b doesn't overflow + if ((a >= 0) == (b >= 0)) return a <= SHRT_MAX/b; // product is positive, so similar to mul2sizes_valid + if (b < 0) return a <= SHRT_MIN / b; // same as a * b >= SHRT_MIN + return a >= SHRT_MIN / b; +} + +// stbi__err - error +// stbi__errpf - error returning pointer to float +// stbi__errpuc - error returning pointer to unsigned char + +#ifdef STBI_NO_FAILURE_STRINGS + #define stbi__err(x,y) 0 +#elif defined(STBI_FAILURE_USERMSG) + #define stbi__err(x,y) stbi__err(y) +#else + #define stbi__err(x,y) stbi__err(x) +#endif + +#define stbi__errpf(x,y) ((float *)(size_t) (stbi__err(x,y)?NULL:NULL)) +#define stbi__errpuc(x,y) ((unsigned char *)(size_t) (stbi__err(x,y)?NULL:NULL)) + +STBIDEF void stbi_image_free(void *retval_from_stbi_load) +{ + STBI_FREE(retval_from_stbi_load); +} + +#ifndef STBI_NO_LINEAR +static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp); +#endif + +#ifndef STBI_NO_HDR +static stbi_uc *stbi__hdr_to_ldr(float *data, int x, int y, int comp); +#endif + +static int stbi__vertically_flip_on_load_global = 0; + +STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip) +{ + stbi__vertically_flip_on_load_global = flag_true_if_should_flip; +} + +#ifndef STBI_THREAD_LOCAL +#define stbi__vertically_flip_on_load stbi__vertically_flip_on_load_global +#else +static STBI_THREAD_LOCAL int stbi__vertically_flip_on_load_local, stbi__vertically_flip_on_load_set; + +STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip) +{ + stbi__vertically_flip_on_load_local = flag_true_if_should_flip; + stbi__vertically_flip_on_load_set = 1; +} + +#define stbi__vertically_flip_on_load (stbi__vertically_flip_on_load_set \ + ? stbi__vertically_flip_on_load_local \ + : stbi__vertically_flip_on_load_global) +#endif // STBI_THREAD_LOCAL + +static void *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc) +{ + memset(ri, 0, sizeof(*ri)); // make sure it's initialized if we add new fields + ri->bits_per_channel = 8; // default is 8 so most paths don't have to be changed + ri->channel_order = STBI_ORDER_RGB; // all current input & output are this, but this is here so we can add BGR order + ri->num_channels = 0; + + // test the formats with a very explicit header first (at least a FOURCC + // or distinctive magic number first) + #ifndef STBI_NO_PNG + if (stbi__png_test(s)) return stbi__png_load(s,x,y,comp,req_comp, ri); + #endif + #ifndef STBI_NO_BMP + if (stbi__bmp_test(s)) return stbi__bmp_load(s,x,y,comp,req_comp, ri); + #endif + #ifndef STBI_NO_GIF + if (stbi__gif_test(s)) return stbi__gif_load(s,x,y,comp,req_comp, ri); + #endif + #ifndef STBI_NO_PSD + if (stbi__psd_test(s)) return stbi__psd_load(s,x,y,comp,req_comp, ri, bpc); + #else + STBI_NOTUSED(bpc); + #endif + #ifndef STBI_NO_PIC + if (stbi__pic_test(s)) return stbi__pic_load(s,x,y,comp,req_comp, ri); + #endif + + // then the formats that can end up attempting to load with just 1 or 2 + // bytes matching expectations; these are prone to false positives, so + // try them later + #ifndef STBI_NO_JPEG + if (stbi__jpeg_test(s)) return stbi__jpeg_load(s,x,y,comp,req_comp, ri); + #endif + #ifndef STBI_NO_PNM + if (stbi__pnm_test(s)) return stbi__pnm_load(s,x,y,comp,req_comp, ri); + #endif + + #ifndef STBI_NO_HDR + if (stbi__hdr_test(s)) { + float *hdr = stbi__hdr_load(s, x,y,comp,req_comp, ri); + return stbi__hdr_to_ldr(hdr, *x, *y, req_comp ? req_comp : *comp); + } + #endif + + #ifndef STBI_NO_TGA + // test tga last because it's a crappy test! + if (stbi__tga_test(s)) + return stbi__tga_load(s,x,y,comp,req_comp, ri); + #endif + + return stbi__errpuc("unknown image type", "Image not of any known type, or corrupt"); +} + +static stbi_uc *stbi__convert_16_to_8(stbi__uint16 *orig, int w, int h, int channels) +{ + int i; + int img_len = w * h * channels; + stbi_uc *reduced; + + reduced = (stbi_uc *) stbi__malloc(img_len); + if (reduced == NULL) return stbi__errpuc("outofmem", "Out of memory"); + + for (i = 0; i < img_len; ++i) + reduced[i] = (stbi_uc)((orig[i] >> 8) & 0xFF); // top half of each byte is sufficient approx of 16->8 bit scaling + + STBI_FREE(orig); + return reduced; +} + +static stbi__uint16 *stbi__convert_8_to_16(stbi_uc *orig, int w, int h, int channels) +{ + int i; + int img_len = w * h * channels; + stbi__uint16 *enlarged; + + enlarged = (stbi__uint16 *) stbi__malloc(img_len*2); + if (enlarged == NULL) return (stbi__uint16 *) stbi__errpuc("outofmem", "Out of memory"); + + for (i = 0; i < img_len; ++i) + enlarged[i] = (stbi__uint16)((orig[i] << 8) + orig[i]); // replicate to high and low byte, maps 0->0, 255->0xffff + + STBI_FREE(orig); + return enlarged; +} + +static void stbi__vertical_flip(void *image, int w, int h, int bytes_per_pixel) +{ + int row; + size_t bytes_per_row = (size_t)w * bytes_per_pixel; + stbi_uc temp[2048]; + stbi_uc *bytes = (stbi_uc *)image; + + for (row = 0; row < (h>>1); row++) { + stbi_uc *row0 = bytes + row*bytes_per_row; + stbi_uc *row1 = bytes + (h - row - 1)*bytes_per_row; + // swap row0 with row1 + size_t bytes_left = bytes_per_row; + while (bytes_left) { + size_t bytes_copy = (bytes_left < sizeof(temp)) ? bytes_left : sizeof(temp); + memcpy(temp, row0, bytes_copy); + memcpy(row0, row1, bytes_copy); + memcpy(row1, temp, bytes_copy); + row0 += bytes_copy; + row1 += bytes_copy; + bytes_left -= bytes_copy; + } + } +} + +#ifndef STBI_NO_GIF +static void stbi__vertical_flip_slices(void *image, int w, int h, int z, int bytes_per_pixel) +{ + int slice; + int slice_size = w * h * bytes_per_pixel; + + stbi_uc *bytes = (stbi_uc *)image; + for (slice = 0; slice < z; ++slice) { + stbi__vertical_flip(bytes, w, h, bytes_per_pixel); + bytes += slice_size; + } +} +#endif + +static unsigned char *stbi__load_and_postprocess_8bit(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + stbi__result_info ri; + void *result = stbi__load_main(s, x, y, comp, req_comp, &ri, 8); + + if (result == NULL) + return NULL; + + // it is the responsibility of the loaders to make sure we get either 8 or 16 bit. + STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16); + + if (ri.bits_per_channel != 8) { + result = stbi__convert_16_to_8((stbi__uint16 *) result, *x, *y, req_comp == 0 ? *comp : req_comp); + ri.bits_per_channel = 8; + } + + // @TODO: move stbi__convert_format to here + + if (stbi__vertically_flip_on_load) { + int channels = req_comp ? req_comp : *comp; + stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi_uc)); + } + + return (unsigned char *) result; +} + +static stbi__uint16 *stbi__load_and_postprocess_16bit(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + stbi__result_info ri; + void *result = stbi__load_main(s, x, y, comp, req_comp, &ri, 16); + + if (result == NULL) + return NULL; + + // it is the responsibility of the loaders to make sure we get either 8 or 16 bit. + STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16); + + if (ri.bits_per_channel != 16) { + result = stbi__convert_8_to_16((stbi_uc *) result, *x, *y, req_comp == 0 ? *comp : req_comp); + ri.bits_per_channel = 16; + } + + // @TODO: move stbi__convert_format16 to here + // @TODO: special case RGB-to-Y (and RGBA-to-YA) for 8-bit-to-16-bit case to keep more precision + + if (stbi__vertically_flip_on_load) { + int channels = req_comp ? req_comp : *comp; + stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi__uint16)); + } + + return (stbi__uint16 *) result; +} + +#if !defined(STBI_NO_HDR) && !defined(STBI_NO_LINEAR) +static void stbi__float_postprocess(float *result, int *x, int *y, int *comp, int req_comp) +{ + if (stbi__vertically_flip_on_load && result != NULL) { + int channels = req_comp ? req_comp : *comp; + stbi__vertical_flip(result, *x, *y, channels * sizeof(float)); + } +} +#endif + +#ifndef STBI_NO_STDIO + +#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) +STBI_EXTERN __declspec(dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char *str, int cbmb, wchar_t *widestr, int cchwide); +STBI_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t *widestr, int cchwide, char *str, int cbmb, const char *defchar, int *used_default); +#endif + +#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) +STBIDEF int stbi_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input) +{ + return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int) bufferlen, NULL, NULL); +} +#endif + +static FILE *stbi__fopen(char const *filename, char const *mode) +{ + FILE *f; +#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) + wchar_t wMode[64]; + wchar_t wFilename[1024]; + if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename)/sizeof(*wFilename))) + return 0; + + if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode)/sizeof(*wMode))) + return 0; + +#if defined(_MSC_VER) && _MSC_VER >= 1400 + if (0 != _wfopen_s(&f, wFilename, wMode)) + f = 0; +#else + f = _wfopen(wFilename, wMode); +#endif + +#elif defined(_MSC_VER) && _MSC_VER >= 1400 + if (0 != fopen_s(&f, filename, mode)) + f=0; +#else + f = fopen(filename, mode); +#endif + return f; +} + + +STBIDEF stbi_uc *stbi_load(char const *filename, int *x, int *y, int *comp, int req_comp) +{ + FILE *f = stbi__fopen(filename, "rb"); + unsigned char *result; + if (!f) return stbi__errpuc("can't fopen", "Unable to open file"); + result = stbi_load_from_file(f,x,y,comp,req_comp); + fclose(f); + return result; +} + +STBIDEF stbi_uc *stbi_load_from_file(FILE *f, int *x, int *y, int *comp, int req_comp) +{ + unsigned char *result; + stbi__context s; + stbi__start_file(&s,f); + result = stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp); + if (result) { + // need to 'unget' all the characters in the IO buffer + fseek(f, - (int) (s.img_buffer_end - s.img_buffer), SEEK_CUR); + } + return result; +} + +STBIDEF stbi__uint16 *stbi_load_from_file_16(FILE *f, int *x, int *y, int *comp, int req_comp) +{ + stbi__uint16 *result; + stbi__context s; + stbi__start_file(&s,f); + result = stbi__load_and_postprocess_16bit(&s,x,y,comp,req_comp); + if (result) { + // need to 'unget' all the characters in the IO buffer + fseek(f, - (int) (s.img_buffer_end - s.img_buffer), SEEK_CUR); + } + return result; +} + +STBIDEF stbi_us *stbi_load_16(char const *filename, int *x, int *y, int *comp, int req_comp) +{ + FILE *f = stbi__fopen(filename, "rb"); + stbi__uint16 *result; + if (!f) return (stbi_us *) stbi__errpuc("can't fopen", "Unable to open file"); + result = stbi_load_from_file_16(f,x,y,comp,req_comp); + fclose(f); + return result; +} + + +#endif //!STBI_NO_STDIO + +STBIDEF stbi_us *stbi_load_16_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels) +{ + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__load_and_postprocess_16bit(&s,x,y,channels_in_file,desired_channels); +} + +STBIDEF stbi_us *stbi_load_16_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *)clbk, user); + return stbi__load_and_postprocess_16bit(&s,x,y,channels_in_file,desired_channels); +} + +STBIDEF stbi_uc *stbi_load_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp); +} + +STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user); + return stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp); +} + +#ifndef STBI_NO_GIF +STBIDEF stbi_uc *stbi_load_gif_from_memory(stbi_uc const *buffer, int len, int **delays, int *x, int *y, int *z, int *comp, int req_comp) +{ + unsigned char *result; + stbi__context s; + stbi__start_mem(&s,buffer,len); + + result = (unsigned char*) stbi__load_gif_main(&s, delays, x, y, z, comp, req_comp); + if (stbi__vertically_flip_on_load) { + stbi__vertical_flip_slices( result, *x, *y, *z, *comp ); + } + + return result; +} +#endif + +#ifndef STBI_NO_LINEAR +static float *stbi__loadf_main(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + unsigned char *data; + #ifndef STBI_NO_HDR + if (stbi__hdr_test(s)) { + stbi__result_info ri; + float *hdr_data = stbi__hdr_load(s,x,y,comp,req_comp, &ri); + if (hdr_data) + stbi__float_postprocess(hdr_data,x,y,comp,req_comp); + return hdr_data; + } + #endif + data = stbi__load_and_postprocess_8bit(s, x, y, comp, req_comp); + if (data) + return stbi__ldr_to_hdr(data, *x, *y, req_comp ? req_comp : *comp); + return stbi__errpf("unknown image type", "Image not of any known type, or corrupt"); +} + +STBIDEF float *stbi_loadf_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__loadf_main(&s,x,y,comp,req_comp); +} + +STBIDEF float *stbi_loadf_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user); + return stbi__loadf_main(&s,x,y,comp,req_comp); +} + +#ifndef STBI_NO_STDIO +STBIDEF float *stbi_loadf(char const *filename, int *x, int *y, int *comp, int req_comp) +{ + float *result; + FILE *f = stbi__fopen(filename, "rb"); + if (!f) return stbi__errpf("can't fopen", "Unable to open file"); + result = stbi_loadf_from_file(f,x,y,comp,req_comp); + fclose(f); + return result; +} + +STBIDEF float *stbi_loadf_from_file(FILE *f, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_file(&s,f); + return stbi__loadf_main(&s,x,y,comp,req_comp); +} +#endif // !STBI_NO_STDIO + +#endif // !STBI_NO_LINEAR + +// these is-hdr-or-not is defined independent of whether STBI_NO_LINEAR is +// defined, for API simplicity; if STBI_NO_LINEAR is defined, it always +// reports false! + +STBIDEF int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len) +{ + #ifndef STBI_NO_HDR + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__hdr_test(&s); + #else + STBI_NOTUSED(buffer); + STBI_NOTUSED(len); + return 0; + #endif +} + +#ifndef STBI_NO_STDIO +STBIDEF int stbi_is_hdr (char const *filename) +{ + FILE *f = stbi__fopen(filename, "rb"); + int result=0; + if (f) { + result = stbi_is_hdr_from_file(f); + fclose(f); + } + return result; +} + +STBIDEF int stbi_is_hdr_from_file(FILE *f) +{ + #ifndef STBI_NO_HDR + long pos = ftell(f); + int res; + stbi__context s; + stbi__start_file(&s,f); + res = stbi__hdr_test(&s); + fseek(f, pos, SEEK_SET); + return res; + #else + STBI_NOTUSED(f); + return 0; + #endif +} +#endif // !STBI_NO_STDIO + +STBIDEF int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user) +{ + #ifndef STBI_NO_HDR + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user); + return stbi__hdr_test(&s); + #else + STBI_NOTUSED(clbk); + STBI_NOTUSED(user); + return 0; + #endif +} + +#ifndef STBI_NO_LINEAR +static float stbi__l2h_gamma=2.2f, stbi__l2h_scale=1.0f; + +STBIDEF void stbi_ldr_to_hdr_gamma(float gamma) { stbi__l2h_gamma = gamma; } +STBIDEF void stbi_ldr_to_hdr_scale(float scale) { stbi__l2h_scale = scale; } +#endif + +static float stbi__h2l_gamma_i=1.0f/2.2f, stbi__h2l_scale_i=1.0f; + +STBIDEF void stbi_hdr_to_ldr_gamma(float gamma) { stbi__h2l_gamma_i = 1/gamma; } +STBIDEF void stbi_hdr_to_ldr_scale(float scale) { stbi__h2l_scale_i = 1/scale; } + + +////////////////////////////////////////////////////////////////////////////// +// +// Common code used by all image loaders +// + +enum +{ + STBI__SCAN_load=0, + STBI__SCAN_type, + STBI__SCAN_header +}; + +static void stbi__refill_buffer(stbi__context *s) +{ + int n = (s->io.read)(s->io_user_data,(char*)s->buffer_start,s->buflen); + s->callback_already_read += (int) (s->img_buffer - s->img_buffer_original); + if (n == 0) { + // at end of file, treat same as if from memory, but need to handle case + // where s->img_buffer isn't pointing to safe memory, e.g. 0-byte file + s->read_from_callbacks = 0; + s->img_buffer = s->buffer_start; + s->img_buffer_end = s->buffer_start+1; + *s->img_buffer = 0; + } else { + s->img_buffer = s->buffer_start; + s->img_buffer_end = s->buffer_start + n; + } +} + +stbi_inline static stbi_uc stbi__get8(stbi__context *s) +{ + if (s->img_buffer < s->img_buffer_end) + return *s->img_buffer++; + if (s->read_from_callbacks) { + stbi__refill_buffer(s); + return *s->img_buffer++; + } + return 0; +} + +#if defined(STBI_NO_JPEG) && defined(STBI_NO_HDR) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM) +// nothing +#else +stbi_inline static int stbi__at_eof(stbi__context *s) +{ + if (s->io.read) { + if (!(s->io.eof)(s->io_user_data)) return 0; + // if feof() is true, check if buffer = end + // special case: we've only got the special 0 character at the end + if (s->read_from_callbacks == 0) return 1; + } + + return s->img_buffer >= s->img_buffer_end; +} +#endif + +#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC) +// nothing +#else +static void stbi__skip(stbi__context *s, int n) +{ + if (n == 0) return; // already there! + if (n < 0) { + s->img_buffer = s->img_buffer_end; + return; + } + if (s->io.read) { + int blen = (int) (s->img_buffer_end - s->img_buffer); + if (blen < n) { + s->img_buffer = s->img_buffer_end; + (s->io.skip)(s->io_user_data, n - blen); + return; + } + } + s->img_buffer += n; +} +#endif + +#if defined(STBI_NO_PNG) && defined(STBI_NO_TGA) && defined(STBI_NO_HDR) && defined(STBI_NO_PNM) +// nothing +#else +static int stbi__getn(stbi__context *s, stbi_uc *buffer, int n) +{ + if (s->io.read) { + int blen = (int) (s->img_buffer_end - s->img_buffer); + if (blen < n) { + int res, count; + + memcpy(buffer, s->img_buffer, blen); + + count = (s->io.read)(s->io_user_data, (char*) buffer + blen, n - blen); + res = (count == (n-blen)); + s->img_buffer = s->img_buffer_end; + return res; + } + } + + if (s->img_buffer+n <= s->img_buffer_end) { + memcpy(buffer, s->img_buffer, n); + s->img_buffer += n; + return 1; + } else + return 0; +} +#endif + +#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC) +// nothing +#else +static int stbi__get16be(stbi__context *s) +{ + int z = stbi__get8(s); + return (z << 8) + stbi__get8(s); +} +#endif + +#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC) +// nothing +#else +static stbi__uint32 stbi__get32be(stbi__context *s) +{ + stbi__uint32 z = stbi__get16be(s); + return (z << 16) + stbi__get16be(s); +} +#endif + +#if defined(STBI_NO_BMP) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) +// nothing +#else +static int stbi__get16le(stbi__context *s) +{ + int z = stbi__get8(s); + return z + (stbi__get8(s) << 8); +} +#endif + +#ifndef STBI_NO_BMP +static stbi__uint32 stbi__get32le(stbi__context *s) +{ + stbi__uint32 z = stbi__get16le(s); + z += (stbi__uint32)stbi__get16le(s) << 16; + return z; +} +#endif + +#define STBI__BYTECAST(x) ((stbi_uc) ((x) & 255)) // truncate int to byte without warnings + +#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM) +// nothing +#else +////////////////////////////////////////////////////////////////////////////// +// +// generic converter from built-in img_n to req_comp +// individual types do this automatically as much as possible (e.g. jpeg +// does all cases internally since it needs to colorspace convert anyway, +// and it never has alpha, so very few cases ). png can automatically +// interleave an alpha=255 channel, but falls back to this for other cases +// +// assume data buffer is malloced, so malloc a new one and free that one +// only failure mode is malloc failing + +static stbi_uc stbi__compute_y(int r, int g, int b) +{ + return (stbi_uc) (((r*77) + (g*150) + (29*b)) >> 8); +} +#endif + +#if defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM) +// nothing +#else +static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int req_comp, unsigned int x, unsigned int y) +{ + int i,j; + unsigned char *good; + + if (req_comp == img_n) return data; + STBI_ASSERT(req_comp >= 1 && req_comp <= 4); + + good = (unsigned char *) stbi__malloc_mad3(req_comp, x, y, 0); + if (good == NULL) { + STBI_FREE(data); + return stbi__errpuc("outofmem", "Out of memory"); + } + + for (j=0; j < (int) y; ++j) { + unsigned char *src = data + j * x * img_n ; + unsigned char *dest = good + j * x * req_comp; + + #define STBI__COMBO(a,b) ((a)*8+(b)) + #define STBI__CASE(a,b) case STBI__COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b) + // convert source image with img_n components to one with req_comp components; + // avoid switch per pixel, so use switch per scanline and massive macros + switch (STBI__COMBO(img_n, req_comp)) { + STBI__CASE(1,2) { dest[0]=src[0]; dest[1]=255; } break; + STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0]; } break; + STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=255; } break; + STBI__CASE(2,1) { dest[0]=src[0]; } break; + STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0]; } break; + STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=src[1]; } break; + STBI__CASE(3,4) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2];dest[3]=255; } break; + STBI__CASE(3,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); } break; + STBI__CASE(3,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); dest[1] = 255; } break; + STBI__CASE(4,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); } break; + STBI__CASE(4,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); dest[1] = src[3]; } break; + STBI__CASE(4,3) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2]; } break; + default: STBI_ASSERT(0); STBI_FREE(data); STBI_FREE(good); return stbi__errpuc("unsupported", "Unsupported format conversion"); + } + #undef STBI__CASE + } + + STBI_FREE(data); + return good; +} +#endif + +#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) +// nothing +#else +static stbi__uint16 stbi__compute_y_16(int r, int g, int b) +{ + return (stbi__uint16) (((r*77) + (g*150) + (29*b)) >> 8); +} +#endif + +#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) +// nothing +#else +static stbi__uint16 *stbi__convert_format16(stbi__uint16 *data, int img_n, int req_comp, unsigned int x, unsigned int y) +{ + int i,j; + stbi__uint16 *good; + + if (req_comp == img_n) return data; + STBI_ASSERT(req_comp >= 1 && req_comp <= 4); + + good = (stbi__uint16 *) stbi__malloc(req_comp * x * y * 2); + if (good == NULL) { + STBI_FREE(data); + return (stbi__uint16 *) stbi__errpuc("outofmem", "Out of memory"); + } + + for (j=0; j < (int) y; ++j) { + stbi__uint16 *src = data + j * x * img_n ; + stbi__uint16 *dest = good + j * x * req_comp; + + #define STBI__COMBO(a,b) ((a)*8+(b)) + #define STBI__CASE(a,b) case STBI__COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b) + // convert source image with img_n components to one with req_comp components; + // avoid switch per pixel, so use switch per scanline and massive macros + switch (STBI__COMBO(img_n, req_comp)) { + STBI__CASE(1,2) { dest[0]=src[0]; dest[1]=0xffff; } break; + STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0]; } break; + STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=0xffff; } break; + STBI__CASE(2,1) { dest[0]=src[0]; } break; + STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0]; } break; + STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=src[1]; } break; + STBI__CASE(3,4) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2];dest[3]=0xffff; } break; + STBI__CASE(3,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); } break; + STBI__CASE(3,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); dest[1] = 0xffff; } break; + STBI__CASE(4,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); } break; + STBI__CASE(4,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); dest[1] = src[3]; } break; + STBI__CASE(4,3) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2]; } break; + default: STBI_ASSERT(0); STBI_FREE(data); STBI_FREE(good); return (stbi__uint16*) stbi__errpuc("unsupported", "Unsupported format conversion"); + } + #undef STBI__CASE + } + + STBI_FREE(data); + return good; +} +#endif + +#ifndef STBI_NO_LINEAR +static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp) +{ + int i,k,n; + float *output; + if (!data) return NULL; + output = (float *) stbi__malloc_mad4(x, y, comp, sizeof(float), 0); + if (output == NULL) { STBI_FREE(data); return stbi__errpf("outofmem", "Out of memory"); } + // compute number of non-alpha components + if (comp & 1) n = comp; else n = comp-1; + for (i=0; i < x*y; ++i) { + for (k=0; k < n; ++k) { + output[i*comp + k] = (float) (pow(data[i*comp+k]/255.0f, stbi__l2h_gamma) * stbi__l2h_scale); + } + } + if (n < comp) { + for (i=0; i < x*y; ++i) { + output[i*comp + n] = data[i*comp + n]/255.0f; + } + } + STBI_FREE(data); + return output; +} +#endif + +#ifndef STBI_NO_HDR +#define stbi__float2int(x) ((int) (x)) +static stbi_uc *stbi__hdr_to_ldr(float *data, int x, int y, int comp) +{ + int i,k,n; + stbi_uc *output; + if (!data) return NULL; + output = (stbi_uc *) stbi__malloc_mad3(x, y, comp, 0); + if (output == NULL) { STBI_FREE(data); return stbi__errpuc("outofmem", "Out of memory"); } + // compute number of non-alpha components + if (comp & 1) n = comp; else n = comp-1; + for (i=0; i < x*y; ++i) { + for (k=0; k < n; ++k) { + float z = (float) pow(data[i*comp+k]*stbi__h2l_scale_i, stbi__h2l_gamma_i) * 255 + 0.5f; + if (z < 0) z = 0; + if (z > 255) z = 255; + output[i*comp + k] = (stbi_uc) stbi__float2int(z); + } + if (k < comp) { + float z = data[i*comp+k] * 255 + 0.5f; + if (z < 0) z = 0; + if (z > 255) z = 255; + output[i*comp + k] = (stbi_uc) stbi__float2int(z); + } + } + STBI_FREE(data); + return output; +} +#endif + +////////////////////////////////////////////////////////////////////////////// +// +// "baseline" JPEG/JFIF decoder +// +// simple implementation +// - doesn't support delayed output of y-dimension +// - simple interface (only one output format: 8-bit interleaved RGB) +// - doesn't try to recover corrupt jpegs +// - doesn't allow partial loading, loading multiple at once +// - still fast on x86 (copying globals into locals doesn't help x86) +// - allocates lots of intermediate memory (full size of all components) +// - non-interleaved case requires this anyway +// - allows good upsampling (see next) +// high-quality +// - upsampled channels are bilinearly interpolated, even across blocks +// - quality integer IDCT derived from IJG's 'slow' +// performance +// - fast huffman; reasonable integer IDCT +// - some SIMD kernels for common paths on targets with SSE2/NEON +// - uses a lot of intermediate memory, could cache poorly + +#ifndef STBI_NO_JPEG + +// huffman decoding acceleration +#define FAST_BITS 9 // larger handles more cases; smaller stomps less cache + +typedef struct +{ + stbi_uc fast[1 << FAST_BITS]; + // weirdly, repacking this into AoS is a 10% speed loss, instead of a win + stbi__uint16 code[256]; + stbi_uc values[256]; + stbi_uc size[257]; + unsigned int maxcode[18]; + int delta[17]; // old 'firstsymbol' - old 'firstcode' +} stbi__huffman; + +typedef struct +{ + stbi__context *s; + stbi__huffman huff_dc[4]; + stbi__huffman huff_ac[4]; + stbi__uint16 dequant[4][64]; + stbi__int16 fast_ac[4][1 << FAST_BITS]; + +// sizes for components, interleaved MCUs + int img_h_max, img_v_max; + int img_mcu_x, img_mcu_y; + int img_mcu_w, img_mcu_h; + +// definition of jpeg image component + struct + { + int id; + int h,v; + int tq; + int hd,ha; + int dc_pred; + + int x,y,w2,h2; + stbi_uc *data; + void *raw_data, *raw_coeff; + stbi_uc *linebuf; + short *coeff; // progressive only + int coeff_w, coeff_h; // number of 8x8 coefficient blocks + } img_comp[4]; + + stbi__uint32 code_buffer; // jpeg entropy-coded buffer + int code_bits; // number of valid bits + unsigned char marker; // marker seen while filling entropy buffer + int nomore; // flag if we saw a marker so must stop + + int progressive; + int spec_start; + int spec_end; + int succ_high; + int succ_low; + int eob_run; + int jfif; + int app14_color_transform; // Adobe APP14 tag + int rgb; + + int scan_n, order[4]; + int restart_interval, todo; + +// kernels + void (*idct_block_kernel)(stbi_uc *out, int out_stride, short data[64]); + void (*YCbCr_to_RGB_kernel)(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step); + stbi_uc *(*resample_row_hv_2_kernel)(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs); +} stbi__jpeg; + +static int stbi__build_huffman(stbi__huffman *h, int *count) +{ + int i,j,k=0; + unsigned int code; + // build size list for each symbol (from JPEG spec) + for (i=0; i < 16; ++i) { + for (j=0; j < count[i]; ++j) { + h->size[k++] = (stbi_uc) (i+1); + if(k >= 257) return stbi__err("bad size list","Corrupt JPEG"); + } + } + h->size[k] = 0; + + // compute actual symbols (from jpeg spec) + code = 0; + k = 0; + for(j=1; j <= 16; ++j) { + // compute delta to add to code to compute symbol id + h->delta[j] = k - code; + if (h->size[k] == j) { + while (h->size[k] == j) + h->code[k++] = (stbi__uint16) (code++); + if (code-1 >= (1u << j)) return stbi__err("bad code lengths","Corrupt JPEG"); + } + // compute largest code + 1 for this size, preshifted as needed later + h->maxcode[j] = code << (16-j); + code <<= 1; + } + h->maxcode[j] = 0xffffffff; + + // build non-spec acceleration table; 255 is flag for not-accelerated + memset(h->fast, 255, 1 << FAST_BITS); + for (i=0; i < k; ++i) { + int s = h->size[i]; + if (s <= FAST_BITS) { + int c = h->code[i] << (FAST_BITS-s); + int m = 1 << (FAST_BITS-s); + for (j=0; j < m; ++j) { + h->fast[c+j] = (stbi_uc) i; + } + } + } + return 1; +} + +// build a table that decodes both magnitude and value of small ACs in +// one go. +static void stbi__build_fast_ac(stbi__int16 *fast_ac, stbi__huffman *h) +{ + int i; + for (i=0; i < (1 << FAST_BITS); ++i) { + stbi_uc fast = h->fast[i]; + fast_ac[i] = 0; + if (fast < 255) { + int rs = h->values[fast]; + int run = (rs >> 4) & 15; + int magbits = rs & 15; + int len = h->size[fast]; + + if (magbits && len + magbits <= FAST_BITS) { + // magnitude code followed by receive_extend code + int k = ((i << len) & ((1 << FAST_BITS) - 1)) >> (FAST_BITS - magbits); + int m = 1 << (magbits - 1); + if (k < m) k += (~0U << magbits) + 1; + // if the result is small enough, we can fit it in fast_ac table + if (k >= -128 && k <= 127) + fast_ac[i] = (stbi__int16) ((k * 256) + (run * 16) + (len + magbits)); + } + } + } +} + +static void stbi__grow_buffer_unsafe(stbi__jpeg *j) +{ + do { + unsigned int b = j->nomore ? 0 : stbi__get8(j->s); + if (b == 0xff) { + int c = stbi__get8(j->s); + while (c == 0xff) c = stbi__get8(j->s); // consume fill bytes + if (c != 0) { + j->marker = (unsigned char) c; + j->nomore = 1; + return; + } + } + j->code_buffer |= b << (24 - j->code_bits); + j->code_bits += 8; + } while (j->code_bits <= 24); +} + +// (1 << n) - 1 +static const stbi__uint32 stbi__bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535}; + +// decode a jpeg huffman value from the bitstream +stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h) +{ + unsigned int temp; + int c,k; + + if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); + + // look at the top FAST_BITS and determine what symbol ID it is, + // if the code is <= FAST_BITS + c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1); + k = h->fast[c]; + if (k < 255) { + int s = h->size[k]; + if (s > j->code_bits) + return -1; + j->code_buffer <<= s; + j->code_bits -= s; + return h->values[k]; + } + + // naive test is to shift the code_buffer down so k bits are + // valid, then test against maxcode. To speed this up, we've + // preshifted maxcode left so that it has (16-k) 0s at the + // end; in other words, regardless of the number of bits, it + // wants to be compared against something shifted to have 16; + // that way we don't need to shift inside the loop. + temp = j->code_buffer >> 16; + for (k=FAST_BITS+1 ; ; ++k) + if (temp < h->maxcode[k]) + break; + if (k == 17) { + // error! code not found + j->code_bits -= 16; + return -1; + } + + if (k > j->code_bits) + return -1; + + // convert the huffman code to the symbol id + c = ((j->code_buffer >> (32 - k)) & stbi__bmask[k]) + h->delta[k]; + if(c < 0 || c >= 256) // symbol id out of bounds! + return -1; + STBI_ASSERT((((j->code_buffer) >> (32 - h->size[c])) & stbi__bmask[h->size[c]]) == h->code[c]); + + // convert the id to a symbol + j->code_bits -= k; + j->code_buffer <<= k; + return h->values[c]; +} + +// bias[n] = (-1<code_bits < n) stbi__grow_buffer_unsafe(j); + if (j->code_bits < n) return 0; // ran out of bits from stream, return 0s intead of continuing + + sgn = j->code_buffer >> 31; // sign bit always in MSB; 0 if MSB clear (positive), 1 if MSB set (negative) + k = stbi_lrot(j->code_buffer, n); + j->code_buffer = k & ~stbi__bmask[n]; + k &= stbi__bmask[n]; + j->code_bits -= n; + return k + (stbi__jbias[n] & (sgn - 1)); +} + +// get some unsigned bits +stbi_inline static int stbi__jpeg_get_bits(stbi__jpeg *j, int n) +{ + unsigned int k; + if (j->code_bits < n) stbi__grow_buffer_unsafe(j); + if (j->code_bits < n) return 0; // ran out of bits from stream, return 0s intead of continuing + k = stbi_lrot(j->code_buffer, n); + j->code_buffer = k & ~stbi__bmask[n]; + k &= stbi__bmask[n]; + j->code_bits -= n; + return k; +} + +stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg *j) +{ + unsigned int k; + if (j->code_bits < 1) stbi__grow_buffer_unsafe(j); + if (j->code_bits < 1) return 0; // ran out of bits from stream, return 0s intead of continuing + k = j->code_buffer; + j->code_buffer <<= 1; + --j->code_bits; + return k & 0x80000000; +} + +// given a value that's at position X in the zigzag stream, +// where does it appear in the 8x8 matrix coded as row-major? +static const stbi_uc stbi__jpeg_dezigzag[64+15] = +{ + 0, 1, 8, 16, 9, 2, 3, 10, + 17, 24, 32, 25, 18, 11, 4, 5, + 12, 19, 26, 33, 40, 48, 41, 34, + 27, 20, 13, 6, 7, 14, 21, 28, + 35, 42, 49, 56, 57, 50, 43, 36, + 29, 22, 15, 23, 30, 37, 44, 51, + 58, 59, 52, 45, 38, 31, 39, 46, + 53, 60, 61, 54, 47, 55, 62, 63, + // let corrupt input sample past end + 63, 63, 63, 63, 63, 63, 63, 63, + 63, 63, 63, 63, 63, 63, 63 +}; + +// decode one 64-entry block-- +static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman *hdc, stbi__huffman *hac, stbi__int16 *fac, int b, stbi__uint16 *dequant) +{ + int diff,dc,k; + int t; + + if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); + t = stbi__jpeg_huff_decode(j, hdc); + if (t < 0 || t > 15) return stbi__err("bad huffman code","Corrupt JPEG"); + + // 0 all the ac values now so we can do it 32-bits at a time + memset(data,0,64*sizeof(data[0])); + + diff = t ? stbi__extend_receive(j, t) : 0; + if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) return stbi__err("bad delta","Corrupt JPEG"); + dc = j->img_comp[b].dc_pred + diff; + j->img_comp[b].dc_pred = dc; + if (!stbi__mul2shorts_valid(dc, dequant[0])) return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + data[0] = (short) (dc * dequant[0]); + + // decode AC components, see JPEG spec + k = 1; + do { + unsigned int zig; + int c,r,s; + if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); + c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1); + r = fac[c]; + if (r) { // fast-AC path + k += (r >> 4) & 15; // run + s = r & 15; // combined length + if (s > j->code_bits) return stbi__err("bad huffman code", "Combined length longer than code bits available"); + j->code_buffer <<= s; + j->code_bits -= s; + // decode into unzigzag'd location + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short) ((r >> 8) * dequant[zig]); + } else { + int rs = stbi__jpeg_huff_decode(j, hac); + if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG"); + s = rs & 15; + r = rs >> 4; + if (s == 0) { + if (rs != 0xf0) break; // end block + k += 16; + } else { + k += r; + // decode into unzigzag'd location + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short) (stbi__extend_receive(j,s) * dequant[zig]); + } + } + } while (k < 64); + return 1; +} + +static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg *j, short data[64], stbi__huffman *hdc, int b) +{ + int diff,dc; + int t; + if (j->spec_end != 0) return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + + if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); + + if (j->succ_high == 0) { + // first scan for DC coefficient, must be first + memset(data,0,64*sizeof(data[0])); // 0 all the ac values now + t = stbi__jpeg_huff_decode(j, hdc); + if (t < 0 || t > 15) return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + diff = t ? stbi__extend_receive(j, t) : 0; + + if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) return stbi__err("bad delta", "Corrupt JPEG"); + dc = j->img_comp[b].dc_pred + diff; + j->img_comp[b].dc_pred = dc; + if (!stbi__mul2shorts_valid(dc, 1 << j->succ_low)) return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + data[0] = (short) (dc * (1 << j->succ_low)); + } else { + // refinement scan for DC coefficient + if (stbi__jpeg_get_bit(j)) + data[0] += (short) (1 << j->succ_low); + } + return 1; +} + +// @OPTIMIZE: store non-zigzagged during the decode passes, +// and only de-zigzag when dequantizing +static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__huffman *hac, stbi__int16 *fac) +{ + int k; + if (j->spec_start == 0) return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + + if (j->succ_high == 0) { + int shift = j->succ_low; + + if (j->eob_run) { + --j->eob_run; + return 1; + } + + k = j->spec_start; + do { + unsigned int zig; + int c,r,s; + if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); + c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1); + r = fac[c]; + if (r) { // fast-AC path + k += (r >> 4) & 15; // run + s = r & 15; // combined length + if (s > j->code_bits) return stbi__err("bad huffman code", "Combined length longer than code bits available"); + j->code_buffer <<= s; + j->code_bits -= s; + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short) ((r >> 8) * (1 << shift)); + } else { + int rs = stbi__jpeg_huff_decode(j, hac); + if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG"); + s = rs & 15; + r = rs >> 4; + if (s == 0) { + if (r < 15) { + j->eob_run = (1 << r); + if (r) + j->eob_run += stbi__jpeg_get_bits(j, r); + --j->eob_run; + break; + } + k += 16; + } else { + k += r; + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short) (stbi__extend_receive(j,s) * (1 << shift)); + } + } + } while (k <= j->spec_end); + } else { + // refinement scan for these AC coefficients + + short bit = (short) (1 << j->succ_low); + + if (j->eob_run) { + --j->eob_run; + for (k = j->spec_start; k <= j->spec_end; ++k) { + short *p = &data[stbi__jpeg_dezigzag[k]]; + if (*p != 0) + if (stbi__jpeg_get_bit(j)) + if ((*p & bit)==0) { + if (*p > 0) + *p += bit; + else + *p -= bit; + } + } + } else { + k = j->spec_start; + do { + int r,s; + int rs = stbi__jpeg_huff_decode(j, hac); // @OPTIMIZE see if we can use the fast path here, advance-by-r is so slow, eh + if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG"); + s = rs & 15; + r = rs >> 4; + if (s == 0) { + if (r < 15) { + j->eob_run = (1 << r) - 1; + if (r) + j->eob_run += stbi__jpeg_get_bits(j, r); + r = 64; // force end of block + } else { + // r=15 s=0 should write 16 0s, so we just do + // a run of 15 0s and then write s (which is 0), + // so we don't have to do anything special here + } + } else { + if (s != 1) return stbi__err("bad huffman code", "Corrupt JPEG"); + // sign bit + if (stbi__jpeg_get_bit(j)) + s = bit; + else + s = -bit; + } + + // advance by r + while (k <= j->spec_end) { + short *p = &data[stbi__jpeg_dezigzag[k++]]; + if (*p != 0) { + if (stbi__jpeg_get_bit(j)) + if ((*p & bit)==0) { + if (*p > 0) + *p += bit; + else + *p -= bit; + } + } else { + if (r == 0) { + *p = (short) s; + break; + } + --r; + } + } + } while (k <= j->spec_end); + } + } + return 1; +} + +// take a -128..127 value and stbi__clamp it and convert to 0..255 +stbi_inline static stbi_uc stbi__clamp(int x) +{ + // trick to use a single test to catch both cases + if ((unsigned int) x > 255) { + if (x < 0) return 0; + if (x > 255) return 255; + } + return (stbi_uc) x; +} + +#define stbi__f2f(x) ((int) (((x) * 4096 + 0.5))) +#define stbi__fsh(x) ((x) * 4096) + +// derived from jidctint -- DCT_ISLOW +#define STBI__IDCT_1D(s0,s1,s2,s3,s4,s5,s6,s7) \ + int t0,t1,t2,t3,p1,p2,p3,p4,p5,x0,x1,x2,x3; \ + p2 = s2; \ + p3 = s6; \ + p1 = (p2+p3) * stbi__f2f(0.5411961f); \ + t2 = p1 + p3*stbi__f2f(-1.847759065f); \ + t3 = p1 + p2*stbi__f2f( 0.765366865f); \ + p2 = s0; \ + p3 = s4; \ + t0 = stbi__fsh(p2+p3); \ + t1 = stbi__fsh(p2-p3); \ + x0 = t0+t3; \ + x3 = t0-t3; \ + x1 = t1+t2; \ + x2 = t1-t2; \ + t0 = s7; \ + t1 = s5; \ + t2 = s3; \ + t3 = s1; \ + p3 = t0+t2; \ + p4 = t1+t3; \ + p1 = t0+t3; \ + p2 = t1+t2; \ + p5 = (p3+p4)*stbi__f2f( 1.175875602f); \ + t0 = t0*stbi__f2f( 0.298631336f); \ + t1 = t1*stbi__f2f( 2.053119869f); \ + t2 = t2*stbi__f2f( 3.072711026f); \ + t3 = t3*stbi__f2f( 1.501321110f); \ + p1 = p5 + p1*stbi__f2f(-0.899976223f); \ + p2 = p5 + p2*stbi__f2f(-2.562915447f); \ + p3 = p3*stbi__f2f(-1.961570560f); \ + p4 = p4*stbi__f2f(-0.390180644f); \ + t3 += p1+p4; \ + t2 += p2+p3; \ + t1 += p2+p4; \ + t0 += p1+p3; + +static void stbi__idct_block(stbi_uc *out, int out_stride, short data[64]) +{ + int i,val[64],*v=val; + stbi_uc *o; + short *d = data; + + // columns + for (i=0; i < 8; ++i,++d, ++v) { + // if all zeroes, shortcut -- this avoids dequantizing 0s and IDCTing + if (d[ 8]==0 && d[16]==0 && d[24]==0 && d[32]==0 + && d[40]==0 && d[48]==0 && d[56]==0) { + // no shortcut 0 seconds + // (1|2|3|4|5|6|7)==0 0 seconds + // all separate -0.047 seconds + // 1 && 2|3 && 4|5 && 6|7: -0.047 seconds + int dcterm = d[0]*4; + v[0] = v[8] = v[16] = v[24] = v[32] = v[40] = v[48] = v[56] = dcterm; + } else { + STBI__IDCT_1D(d[ 0],d[ 8],d[16],d[24],d[32],d[40],d[48],d[56]) + // constants scaled things up by 1<<12; let's bring them back + // down, but keep 2 extra bits of precision + x0 += 512; x1 += 512; x2 += 512; x3 += 512; + v[ 0] = (x0+t3) >> 10; + v[56] = (x0-t3) >> 10; + v[ 8] = (x1+t2) >> 10; + v[48] = (x1-t2) >> 10; + v[16] = (x2+t1) >> 10; + v[40] = (x2-t1) >> 10; + v[24] = (x3+t0) >> 10; + v[32] = (x3-t0) >> 10; + } + } + + for (i=0, v=val, o=out; i < 8; ++i,v+=8,o+=out_stride) { + // no fast case since the first 1D IDCT spread components out + STBI__IDCT_1D(v[0],v[1],v[2],v[3],v[4],v[5],v[6],v[7]) + // constants scaled things up by 1<<12, plus we had 1<<2 from first + // loop, plus horizontal and vertical each scale by sqrt(8) so together + // we've got an extra 1<<3, so 1<<17 total we need to remove. + // so we want to round that, which means adding 0.5 * 1<<17, + // aka 65536. Also, we'll end up with -128 to 127 that we want + // to encode as 0..255 by adding 128, so we'll add that before the shift + x0 += 65536 + (128<<17); + x1 += 65536 + (128<<17); + x2 += 65536 + (128<<17); + x3 += 65536 + (128<<17); + // tried computing the shifts into temps, or'ing the temps to see + // if any were out of range, but that was slower + o[0] = stbi__clamp((x0+t3) >> 17); + o[7] = stbi__clamp((x0-t3) >> 17); + o[1] = stbi__clamp((x1+t2) >> 17); + o[6] = stbi__clamp((x1-t2) >> 17); + o[2] = stbi__clamp((x2+t1) >> 17); + o[5] = stbi__clamp((x2-t1) >> 17); + o[3] = stbi__clamp((x3+t0) >> 17); + o[4] = stbi__clamp((x3-t0) >> 17); + } +} + +#ifdef STBI_SSE2 +// sse2 integer IDCT. not the fastest possible implementation but it +// produces bit-identical results to the generic C version so it's +// fully "transparent". +static void stbi__idct_simd(stbi_uc *out, int out_stride, short data[64]) +{ + // This is constructed to match our regular (generic) integer IDCT exactly. + __m128i row0, row1, row2, row3, row4, row5, row6, row7; + __m128i tmp; + + // dot product constant: even elems=x, odd elems=y + #define dct_const(x,y) _mm_setr_epi16((x),(y),(x),(y),(x),(y),(x),(y)) + + // out(0) = c0[even]*x + c0[odd]*y (c0, x, y 16-bit, out 32-bit) + // out(1) = c1[even]*x + c1[odd]*y + #define dct_rot(out0,out1, x,y,c0,c1) \ + __m128i c0##lo = _mm_unpacklo_epi16((x),(y)); \ + __m128i c0##hi = _mm_unpackhi_epi16((x),(y)); \ + __m128i out0##_l = _mm_madd_epi16(c0##lo, c0); \ + __m128i out0##_h = _mm_madd_epi16(c0##hi, c0); \ + __m128i out1##_l = _mm_madd_epi16(c0##lo, c1); \ + __m128i out1##_h = _mm_madd_epi16(c0##hi, c1) + + // out = in << 12 (in 16-bit, out 32-bit) + #define dct_widen(out, in) \ + __m128i out##_l = _mm_srai_epi32(_mm_unpacklo_epi16(_mm_setzero_si128(), (in)), 4); \ + __m128i out##_h = _mm_srai_epi32(_mm_unpackhi_epi16(_mm_setzero_si128(), (in)), 4) + + // wide add + #define dct_wadd(out, a, b) \ + __m128i out##_l = _mm_add_epi32(a##_l, b##_l); \ + __m128i out##_h = _mm_add_epi32(a##_h, b##_h) + + // wide sub + #define dct_wsub(out, a, b) \ + __m128i out##_l = _mm_sub_epi32(a##_l, b##_l); \ + __m128i out##_h = _mm_sub_epi32(a##_h, b##_h) + + // butterfly a/b, add bias, then shift by "s" and pack + #define dct_bfly32o(out0, out1, a,b,bias,s) \ + { \ + __m128i abiased_l = _mm_add_epi32(a##_l, bias); \ + __m128i abiased_h = _mm_add_epi32(a##_h, bias); \ + dct_wadd(sum, abiased, b); \ + dct_wsub(dif, abiased, b); \ + out0 = _mm_packs_epi32(_mm_srai_epi32(sum_l, s), _mm_srai_epi32(sum_h, s)); \ + out1 = _mm_packs_epi32(_mm_srai_epi32(dif_l, s), _mm_srai_epi32(dif_h, s)); \ + } + + // 8-bit interleave step (for transposes) + #define dct_interleave8(a, b) \ + tmp = a; \ + a = _mm_unpacklo_epi8(a, b); \ + b = _mm_unpackhi_epi8(tmp, b) + + // 16-bit interleave step (for transposes) + #define dct_interleave16(a, b) \ + tmp = a; \ + a = _mm_unpacklo_epi16(a, b); \ + b = _mm_unpackhi_epi16(tmp, b) + + #define dct_pass(bias,shift) \ + { \ + /* even part */ \ + dct_rot(t2e,t3e, row2,row6, rot0_0,rot0_1); \ + __m128i sum04 = _mm_add_epi16(row0, row4); \ + __m128i dif04 = _mm_sub_epi16(row0, row4); \ + dct_widen(t0e, sum04); \ + dct_widen(t1e, dif04); \ + dct_wadd(x0, t0e, t3e); \ + dct_wsub(x3, t0e, t3e); \ + dct_wadd(x1, t1e, t2e); \ + dct_wsub(x2, t1e, t2e); \ + /* odd part */ \ + dct_rot(y0o,y2o, row7,row3, rot2_0,rot2_1); \ + dct_rot(y1o,y3o, row5,row1, rot3_0,rot3_1); \ + __m128i sum17 = _mm_add_epi16(row1, row7); \ + __m128i sum35 = _mm_add_epi16(row3, row5); \ + dct_rot(y4o,y5o, sum17,sum35, rot1_0,rot1_1); \ + dct_wadd(x4, y0o, y4o); \ + dct_wadd(x5, y1o, y5o); \ + dct_wadd(x6, y2o, y5o); \ + dct_wadd(x7, y3o, y4o); \ + dct_bfly32o(row0,row7, x0,x7,bias,shift); \ + dct_bfly32o(row1,row6, x1,x6,bias,shift); \ + dct_bfly32o(row2,row5, x2,x5,bias,shift); \ + dct_bfly32o(row3,row4, x3,x4,bias,shift); \ + } + + __m128i rot0_0 = dct_const(stbi__f2f(0.5411961f), stbi__f2f(0.5411961f) + stbi__f2f(-1.847759065f)); + __m128i rot0_1 = dct_const(stbi__f2f(0.5411961f) + stbi__f2f( 0.765366865f), stbi__f2f(0.5411961f)); + __m128i rot1_0 = dct_const(stbi__f2f(1.175875602f) + stbi__f2f(-0.899976223f), stbi__f2f(1.175875602f)); + __m128i rot1_1 = dct_const(stbi__f2f(1.175875602f), stbi__f2f(1.175875602f) + stbi__f2f(-2.562915447f)); + __m128i rot2_0 = dct_const(stbi__f2f(-1.961570560f) + stbi__f2f( 0.298631336f), stbi__f2f(-1.961570560f)); + __m128i rot2_1 = dct_const(stbi__f2f(-1.961570560f), stbi__f2f(-1.961570560f) + stbi__f2f( 3.072711026f)); + __m128i rot3_0 = dct_const(stbi__f2f(-0.390180644f) + stbi__f2f( 2.053119869f), stbi__f2f(-0.390180644f)); + __m128i rot3_1 = dct_const(stbi__f2f(-0.390180644f), stbi__f2f(-0.390180644f) + stbi__f2f( 1.501321110f)); + + // rounding biases in column/row passes, see stbi__idct_block for explanation. + __m128i bias_0 = _mm_set1_epi32(512); + __m128i bias_1 = _mm_set1_epi32(65536 + (128<<17)); + + // load + row0 = _mm_load_si128((const __m128i *) (data + 0*8)); + row1 = _mm_load_si128((const __m128i *) (data + 1*8)); + row2 = _mm_load_si128((const __m128i *) (data + 2*8)); + row3 = _mm_load_si128((const __m128i *) (data + 3*8)); + row4 = _mm_load_si128((const __m128i *) (data + 4*8)); + row5 = _mm_load_si128((const __m128i *) (data + 5*8)); + row6 = _mm_load_si128((const __m128i *) (data + 6*8)); + row7 = _mm_load_si128((const __m128i *) (data + 7*8)); + + // column pass + dct_pass(bias_0, 10); + + { + // 16bit 8x8 transpose pass 1 + dct_interleave16(row0, row4); + dct_interleave16(row1, row5); + dct_interleave16(row2, row6); + dct_interleave16(row3, row7); + + // transpose pass 2 + dct_interleave16(row0, row2); + dct_interleave16(row1, row3); + dct_interleave16(row4, row6); + dct_interleave16(row5, row7); + + // transpose pass 3 + dct_interleave16(row0, row1); + dct_interleave16(row2, row3); + dct_interleave16(row4, row5); + dct_interleave16(row6, row7); + } + + // row pass + dct_pass(bias_1, 17); + + { + // pack + __m128i p0 = _mm_packus_epi16(row0, row1); // a0a1a2a3...a7b0b1b2b3...b7 + __m128i p1 = _mm_packus_epi16(row2, row3); + __m128i p2 = _mm_packus_epi16(row4, row5); + __m128i p3 = _mm_packus_epi16(row6, row7); + + // 8bit 8x8 transpose pass 1 + dct_interleave8(p0, p2); // a0e0a1e1... + dct_interleave8(p1, p3); // c0g0c1g1... + + // transpose pass 2 + dct_interleave8(p0, p1); // a0c0e0g0... + dct_interleave8(p2, p3); // b0d0f0h0... + + // transpose pass 3 + dct_interleave8(p0, p2); // a0b0c0d0... + dct_interleave8(p1, p3); // a4b4c4d4... + + // store + _mm_storel_epi64((__m128i *) out, p0); out += out_stride; + _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p0, 0x4e)); out += out_stride; + _mm_storel_epi64((__m128i *) out, p2); out += out_stride; + _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p2, 0x4e)); out += out_stride; + _mm_storel_epi64((__m128i *) out, p1); out += out_stride; + _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p1, 0x4e)); out += out_stride; + _mm_storel_epi64((__m128i *) out, p3); out += out_stride; + _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p3, 0x4e)); + } + +#undef dct_const +#undef dct_rot +#undef dct_widen +#undef dct_wadd +#undef dct_wsub +#undef dct_bfly32o +#undef dct_interleave8 +#undef dct_interleave16 +#undef dct_pass +} + +#endif // STBI_SSE2 + +#ifdef STBI_NEON + +// NEON integer IDCT. should produce bit-identical +// results to the generic C version. +static void stbi__idct_simd(stbi_uc *out, int out_stride, short data[64]) +{ + int16x8_t row0, row1, row2, row3, row4, row5, row6, row7; + + int16x4_t rot0_0 = vdup_n_s16(stbi__f2f(0.5411961f)); + int16x4_t rot0_1 = vdup_n_s16(stbi__f2f(-1.847759065f)); + int16x4_t rot0_2 = vdup_n_s16(stbi__f2f( 0.765366865f)); + int16x4_t rot1_0 = vdup_n_s16(stbi__f2f( 1.175875602f)); + int16x4_t rot1_1 = vdup_n_s16(stbi__f2f(-0.899976223f)); + int16x4_t rot1_2 = vdup_n_s16(stbi__f2f(-2.562915447f)); + int16x4_t rot2_0 = vdup_n_s16(stbi__f2f(-1.961570560f)); + int16x4_t rot2_1 = vdup_n_s16(stbi__f2f(-0.390180644f)); + int16x4_t rot3_0 = vdup_n_s16(stbi__f2f( 0.298631336f)); + int16x4_t rot3_1 = vdup_n_s16(stbi__f2f( 2.053119869f)); + int16x4_t rot3_2 = vdup_n_s16(stbi__f2f( 3.072711026f)); + int16x4_t rot3_3 = vdup_n_s16(stbi__f2f( 1.501321110f)); + +#define dct_long_mul(out, inq, coeff) \ + int32x4_t out##_l = vmull_s16(vget_low_s16(inq), coeff); \ + int32x4_t out##_h = vmull_s16(vget_high_s16(inq), coeff) + +#define dct_long_mac(out, acc, inq, coeff) \ + int32x4_t out##_l = vmlal_s16(acc##_l, vget_low_s16(inq), coeff); \ + int32x4_t out##_h = vmlal_s16(acc##_h, vget_high_s16(inq), coeff) + +#define dct_widen(out, inq) \ + int32x4_t out##_l = vshll_n_s16(vget_low_s16(inq), 12); \ + int32x4_t out##_h = vshll_n_s16(vget_high_s16(inq), 12) + +// wide add +#define dct_wadd(out, a, b) \ + int32x4_t out##_l = vaddq_s32(a##_l, b##_l); \ + int32x4_t out##_h = vaddq_s32(a##_h, b##_h) + +// wide sub +#define dct_wsub(out, a, b) \ + int32x4_t out##_l = vsubq_s32(a##_l, b##_l); \ + int32x4_t out##_h = vsubq_s32(a##_h, b##_h) + +// butterfly a/b, then shift using "shiftop" by "s" and pack +#define dct_bfly32o(out0,out1, a,b,shiftop,s) \ + { \ + dct_wadd(sum, a, b); \ + dct_wsub(dif, a, b); \ + out0 = vcombine_s16(shiftop(sum_l, s), shiftop(sum_h, s)); \ + out1 = vcombine_s16(shiftop(dif_l, s), shiftop(dif_h, s)); \ + } + +#define dct_pass(shiftop, shift) \ + { \ + /* even part */ \ + int16x8_t sum26 = vaddq_s16(row2, row6); \ + dct_long_mul(p1e, sum26, rot0_0); \ + dct_long_mac(t2e, p1e, row6, rot0_1); \ + dct_long_mac(t3e, p1e, row2, rot0_2); \ + int16x8_t sum04 = vaddq_s16(row0, row4); \ + int16x8_t dif04 = vsubq_s16(row0, row4); \ + dct_widen(t0e, sum04); \ + dct_widen(t1e, dif04); \ + dct_wadd(x0, t0e, t3e); \ + dct_wsub(x3, t0e, t3e); \ + dct_wadd(x1, t1e, t2e); \ + dct_wsub(x2, t1e, t2e); \ + /* odd part */ \ + int16x8_t sum15 = vaddq_s16(row1, row5); \ + int16x8_t sum17 = vaddq_s16(row1, row7); \ + int16x8_t sum35 = vaddq_s16(row3, row5); \ + int16x8_t sum37 = vaddq_s16(row3, row7); \ + int16x8_t sumodd = vaddq_s16(sum17, sum35); \ + dct_long_mul(p5o, sumodd, rot1_0); \ + dct_long_mac(p1o, p5o, sum17, rot1_1); \ + dct_long_mac(p2o, p5o, sum35, rot1_2); \ + dct_long_mul(p3o, sum37, rot2_0); \ + dct_long_mul(p4o, sum15, rot2_1); \ + dct_wadd(sump13o, p1o, p3o); \ + dct_wadd(sump24o, p2o, p4o); \ + dct_wadd(sump23o, p2o, p3o); \ + dct_wadd(sump14o, p1o, p4o); \ + dct_long_mac(x4, sump13o, row7, rot3_0); \ + dct_long_mac(x5, sump24o, row5, rot3_1); \ + dct_long_mac(x6, sump23o, row3, rot3_2); \ + dct_long_mac(x7, sump14o, row1, rot3_3); \ + dct_bfly32o(row0,row7, x0,x7,shiftop,shift); \ + dct_bfly32o(row1,row6, x1,x6,shiftop,shift); \ + dct_bfly32o(row2,row5, x2,x5,shiftop,shift); \ + dct_bfly32o(row3,row4, x3,x4,shiftop,shift); \ + } + + // load + row0 = vld1q_s16(data + 0*8); + row1 = vld1q_s16(data + 1*8); + row2 = vld1q_s16(data + 2*8); + row3 = vld1q_s16(data + 3*8); + row4 = vld1q_s16(data + 4*8); + row5 = vld1q_s16(data + 5*8); + row6 = vld1q_s16(data + 6*8); + row7 = vld1q_s16(data + 7*8); + + // add DC bias + row0 = vaddq_s16(row0, vsetq_lane_s16(1024, vdupq_n_s16(0), 0)); + + // column pass + dct_pass(vrshrn_n_s32, 10); + + // 16bit 8x8 transpose + { +// these three map to a single VTRN.16, VTRN.32, and VSWP, respectively. +// whether compilers actually get this is another story, sadly. +#define dct_trn16(x, y) { int16x8x2_t t = vtrnq_s16(x, y); x = t.val[0]; y = t.val[1]; } +#define dct_trn32(x, y) { int32x4x2_t t = vtrnq_s32(vreinterpretq_s32_s16(x), vreinterpretq_s32_s16(y)); x = vreinterpretq_s16_s32(t.val[0]); y = vreinterpretq_s16_s32(t.val[1]); } +#define dct_trn64(x, y) { int16x8_t x0 = x; int16x8_t y0 = y; x = vcombine_s16(vget_low_s16(x0), vget_low_s16(y0)); y = vcombine_s16(vget_high_s16(x0), vget_high_s16(y0)); } + + // pass 1 + dct_trn16(row0, row1); // a0b0a2b2a4b4a6b6 + dct_trn16(row2, row3); + dct_trn16(row4, row5); + dct_trn16(row6, row7); + + // pass 2 + dct_trn32(row0, row2); // a0b0c0d0a4b4c4d4 + dct_trn32(row1, row3); + dct_trn32(row4, row6); + dct_trn32(row5, row7); + + // pass 3 + dct_trn64(row0, row4); // a0b0c0d0e0f0g0h0 + dct_trn64(row1, row5); + dct_trn64(row2, row6); + dct_trn64(row3, row7); + +#undef dct_trn16 +#undef dct_trn32 +#undef dct_trn64 + } + + // row pass + // vrshrn_n_s32 only supports shifts up to 16, we need + // 17. so do a non-rounding shift of 16 first then follow + // up with a rounding shift by 1. + dct_pass(vshrn_n_s32, 16); + + { + // pack and round + uint8x8_t p0 = vqrshrun_n_s16(row0, 1); + uint8x8_t p1 = vqrshrun_n_s16(row1, 1); + uint8x8_t p2 = vqrshrun_n_s16(row2, 1); + uint8x8_t p3 = vqrshrun_n_s16(row3, 1); + uint8x8_t p4 = vqrshrun_n_s16(row4, 1); + uint8x8_t p5 = vqrshrun_n_s16(row5, 1); + uint8x8_t p6 = vqrshrun_n_s16(row6, 1); + uint8x8_t p7 = vqrshrun_n_s16(row7, 1); + + // again, these can translate into one instruction, but often don't. +#define dct_trn8_8(x, y) { uint8x8x2_t t = vtrn_u8(x, y); x = t.val[0]; y = t.val[1]; } +#define dct_trn8_16(x, y) { uint16x4x2_t t = vtrn_u16(vreinterpret_u16_u8(x), vreinterpret_u16_u8(y)); x = vreinterpret_u8_u16(t.val[0]); y = vreinterpret_u8_u16(t.val[1]); } +#define dct_trn8_32(x, y) { uint32x2x2_t t = vtrn_u32(vreinterpret_u32_u8(x), vreinterpret_u32_u8(y)); x = vreinterpret_u8_u32(t.val[0]); y = vreinterpret_u8_u32(t.val[1]); } + + // sadly can't use interleaved stores here since we only write + // 8 bytes to each scan line! + + // 8x8 8-bit transpose pass 1 + dct_trn8_8(p0, p1); + dct_trn8_8(p2, p3); + dct_trn8_8(p4, p5); + dct_trn8_8(p6, p7); + + // pass 2 + dct_trn8_16(p0, p2); + dct_trn8_16(p1, p3); + dct_trn8_16(p4, p6); + dct_trn8_16(p5, p7); + + // pass 3 + dct_trn8_32(p0, p4); + dct_trn8_32(p1, p5); + dct_trn8_32(p2, p6); + dct_trn8_32(p3, p7); + + // store + vst1_u8(out, p0); out += out_stride; + vst1_u8(out, p1); out += out_stride; + vst1_u8(out, p2); out += out_stride; + vst1_u8(out, p3); out += out_stride; + vst1_u8(out, p4); out += out_stride; + vst1_u8(out, p5); out += out_stride; + vst1_u8(out, p6); out += out_stride; + vst1_u8(out, p7); + +#undef dct_trn8_8 +#undef dct_trn8_16 +#undef dct_trn8_32 + } + +#undef dct_long_mul +#undef dct_long_mac +#undef dct_widen +#undef dct_wadd +#undef dct_wsub +#undef dct_bfly32o +#undef dct_pass +} + +#endif // STBI_NEON + +#define STBI__MARKER_none 0xff +// if there's a pending marker from the entropy stream, return that +// otherwise, fetch from the stream and get a marker. if there's no +// marker, return 0xff, which is never a valid marker value +static stbi_uc stbi__get_marker(stbi__jpeg *j) +{ + stbi_uc x; + if (j->marker != STBI__MARKER_none) { x = j->marker; j->marker = STBI__MARKER_none; return x; } + x = stbi__get8(j->s); + if (x != 0xff) return STBI__MARKER_none; + while (x == 0xff) + x = stbi__get8(j->s); // consume repeated 0xff fill bytes + return x; +} + +// in each scan, we'll have scan_n components, and the order +// of the components is specified by order[] +#define STBI__RESTART(x) ((x) >= 0xd0 && (x) <= 0xd7) + +// after a restart interval, stbi__jpeg_reset the entropy decoder and +// the dc prediction +static void stbi__jpeg_reset(stbi__jpeg *j) +{ + j->code_bits = 0; + j->code_buffer = 0; + j->nomore = 0; + j->img_comp[0].dc_pred = j->img_comp[1].dc_pred = j->img_comp[2].dc_pred = j->img_comp[3].dc_pred = 0; + j->marker = STBI__MARKER_none; + j->todo = j->restart_interval ? j->restart_interval : 0x7fffffff; + j->eob_run = 0; + // no more than 1<<31 MCUs if no restart_interal? that's plenty safe, + // since we don't even allow 1<<30 pixels +} + +static int stbi__parse_entropy_coded_data(stbi__jpeg *z) +{ + stbi__jpeg_reset(z); + if (!z->progressive) { + if (z->scan_n == 1) { + int i,j; + STBI_SIMD_ALIGN(short, data[64]); + int n = z->order[0]; + // non-interleaved data, we just need to process one block at a time, + // in trivial scanline order + // number of blocks to do just depends on how many actual "pixels" this + // component has, independent of interleaved MCU blocking and such + int w = (z->img_comp[n].x+7) >> 3; + int h = (z->img_comp[n].y+7) >> 3; + for (j=0; j < h; ++j) { + for (i=0; i < w; ++i) { + int ha = z->img_comp[n].ha; + if (!stbi__jpeg_decode_block(z, data, z->huff_dc+z->img_comp[n].hd, z->huff_ac+ha, z->fast_ac[ha], n, z->dequant[z->img_comp[n].tq])) return 0; + z->idct_block_kernel(z->img_comp[n].data+z->img_comp[n].w2*j*8+i*8, z->img_comp[n].w2, data); + // every data block is an MCU, so countdown the restart interval + if (--z->todo <= 0) { + if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); + // if it's NOT a restart, then just bail, so we get corrupt data + // rather than no data + if (!STBI__RESTART(z->marker)) return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } else { // interleaved + int i,j,k,x,y; + STBI_SIMD_ALIGN(short, data[64]); + for (j=0; j < z->img_mcu_y; ++j) { + for (i=0; i < z->img_mcu_x; ++i) { + // scan an interleaved mcu... process scan_n components in order + for (k=0; k < z->scan_n; ++k) { + int n = z->order[k]; + // scan out an mcu's worth of this component; that's just determined + // by the basic H and V specified for the component + for (y=0; y < z->img_comp[n].v; ++y) { + for (x=0; x < z->img_comp[n].h; ++x) { + int x2 = (i*z->img_comp[n].h + x)*8; + int y2 = (j*z->img_comp[n].v + y)*8; + int ha = z->img_comp[n].ha; + if (!stbi__jpeg_decode_block(z, data, z->huff_dc+z->img_comp[n].hd, z->huff_ac+ha, z->fast_ac[ha], n, z->dequant[z->img_comp[n].tq])) return 0; + z->idct_block_kernel(z->img_comp[n].data+z->img_comp[n].w2*y2+x2, z->img_comp[n].w2, data); + } + } + } + // after all interleaved components, that's an interleaved MCU, + // so now count down the restart interval + if (--z->todo <= 0) { + if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); + if (!STBI__RESTART(z->marker)) return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } + } else { + if (z->scan_n == 1) { + int i,j; + int n = z->order[0]; + // non-interleaved data, we just need to process one block at a time, + // in trivial scanline order + // number of blocks to do just depends on how many actual "pixels" this + // component has, independent of interleaved MCU blocking and such + int w = (z->img_comp[n].x+7) >> 3; + int h = (z->img_comp[n].y+7) >> 3; + for (j=0; j < h; ++j) { + for (i=0; i < w; ++i) { + short *data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w); + if (z->spec_start == 0) { + if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n)) + return 0; + } else { + int ha = z->img_comp[n].ha; + if (!stbi__jpeg_decode_block_prog_ac(z, data, &z->huff_ac[ha], z->fast_ac[ha])) + return 0; + } + // every data block is an MCU, so countdown the restart interval + if (--z->todo <= 0) { + if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); + if (!STBI__RESTART(z->marker)) return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } else { // interleaved + int i,j,k,x,y; + for (j=0; j < z->img_mcu_y; ++j) { + for (i=0; i < z->img_mcu_x; ++i) { + // scan an interleaved mcu... process scan_n components in order + for (k=0; k < z->scan_n; ++k) { + int n = z->order[k]; + // scan out an mcu's worth of this component; that's just determined + // by the basic H and V specified for the component + for (y=0; y < z->img_comp[n].v; ++y) { + for (x=0; x < z->img_comp[n].h; ++x) { + int x2 = (i*z->img_comp[n].h + x); + int y2 = (j*z->img_comp[n].v + y); + short *data = z->img_comp[n].coeff + 64 * (x2 + y2 * z->img_comp[n].coeff_w); + if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n)) + return 0; + } + } + } + // after all interleaved components, that's an interleaved MCU, + // so now count down the restart interval + if (--z->todo <= 0) { + if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); + if (!STBI__RESTART(z->marker)) return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } + } +} + +static void stbi__jpeg_dequantize(short *data, stbi__uint16 *dequant) +{ + int i; + for (i=0; i < 64; ++i) + data[i] *= dequant[i]; +} + +static void stbi__jpeg_finish(stbi__jpeg *z) +{ + if (z->progressive) { + // dequantize and idct the data + int i,j,n; + for (n=0; n < z->s->img_n; ++n) { + int w = (z->img_comp[n].x+7) >> 3; + int h = (z->img_comp[n].y+7) >> 3; + for (j=0; j < h; ++j) { + for (i=0; i < w; ++i) { + short *data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w); + stbi__jpeg_dequantize(data, z->dequant[z->img_comp[n].tq]); + z->idct_block_kernel(z->img_comp[n].data+z->img_comp[n].w2*j*8+i*8, z->img_comp[n].w2, data); + } + } + } + } +} + +static int stbi__process_marker(stbi__jpeg *z, int m) +{ + int L; + switch (m) { + case STBI__MARKER_none: // no marker found + return stbi__err("expected marker","Corrupt JPEG"); + + case 0xDD: // DRI - specify restart interval + if (stbi__get16be(z->s) != 4) return stbi__err("bad DRI len","Corrupt JPEG"); + z->restart_interval = stbi__get16be(z->s); + return 1; + + case 0xDB: // DQT - define quantization table + L = stbi__get16be(z->s)-2; + while (L > 0) { + int q = stbi__get8(z->s); + int p = q >> 4, sixteen = (p != 0); + int t = q & 15,i; + if (p != 0 && p != 1) return stbi__err("bad DQT type","Corrupt JPEG"); + if (t > 3) return stbi__err("bad DQT table","Corrupt JPEG"); + + for (i=0; i < 64; ++i) + z->dequant[t][stbi__jpeg_dezigzag[i]] = (stbi__uint16)(sixteen ? stbi__get16be(z->s) : stbi__get8(z->s)); + L -= (sixteen ? 129 : 65); + } + return L==0; + + case 0xC4: // DHT - define huffman table + L = stbi__get16be(z->s)-2; + while (L > 0) { + stbi_uc *v; + int sizes[16],i,n=0; + int q = stbi__get8(z->s); + int tc = q >> 4; + int th = q & 15; + if (tc > 1 || th > 3) return stbi__err("bad DHT header","Corrupt JPEG"); + for (i=0; i < 16; ++i) { + sizes[i] = stbi__get8(z->s); + n += sizes[i]; + } + if(n > 256) return stbi__err("bad DHT header","Corrupt JPEG"); // Loop over i < n would write past end of values! + L -= 17; + if (tc == 0) { + if (!stbi__build_huffman(z->huff_dc+th, sizes)) return 0; + v = z->huff_dc[th].values; + } else { + if (!stbi__build_huffman(z->huff_ac+th, sizes)) return 0; + v = z->huff_ac[th].values; + } + for (i=0; i < n; ++i) + v[i] = stbi__get8(z->s); + if (tc != 0) + stbi__build_fast_ac(z->fast_ac[th], z->huff_ac + th); + L -= n; + } + return L==0; + } + + // check for comment block or APP blocks + if ((m >= 0xE0 && m <= 0xEF) || m == 0xFE) { + L = stbi__get16be(z->s); + if (L < 2) { + if (m == 0xFE) + return stbi__err("bad COM len","Corrupt JPEG"); + else + return stbi__err("bad APP len","Corrupt JPEG"); + } + L -= 2; + + if (m == 0xE0 && L >= 5) { // JFIF APP0 segment + static const unsigned char tag[5] = {'J','F','I','F','\0'}; + int ok = 1; + int i; + for (i=0; i < 5; ++i) + if (stbi__get8(z->s) != tag[i]) + ok = 0; + L -= 5; + if (ok) + z->jfif = 1; + } else if (m == 0xEE && L >= 12) { // Adobe APP14 segment + static const unsigned char tag[6] = {'A','d','o','b','e','\0'}; + int ok = 1; + int i; + for (i=0; i < 6; ++i) + if (stbi__get8(z->s) != tag[i]) + ok = 0; + L -= 6; + if (ok) { + stbi__get8(z->s); // version + stbi__get16be(z->s); // flags0 + stbi__get16be(z->s); // flags1 + z->app14_color_transform = stbi__get8(z->s); // color transform + L -= 6; + } + } + + stbi__skip(z->s, L); + return 1; + } + + return stbi__err("unknown marker","Corrupt JPEG"); +} + +// after we see SOS +static int stbi__process_scan_header(stbi__jpeg *z) +{ + int i; + int Ls = stbi__get16be(z->s); + z->scan_n = stbi__get8(z->s); + if (z->scan_n < 1 || z->scan_n > 4 || z->scan_n > (int) z->s->img_n) return stbi__err("bad SOS component count","Corrupt JPEG"); + if (Ls != 6+2*z->scan_n) return stbi__err("bad SOS len","Corrupt JPEG"); + for (i=0; i < z->scan_n; ++i) { + int id = stbi__get8(z->s), which; + int q = stbi__get8(z->s); + for (which = 0; which < z->s->img_n; ++which) + if (z->img_comp[which].id == id) + break; + if (which == z->s->img_n) return 0; // no match + z->img_comp[which].hd = q >> 4; if (z->img_comp[which].hd > 3) return stbi__err("bad DC huff","Corrupt JPEG"); + z->img_comp[which].ha = q & 15; if (z->img_comp[which].ha > 3) return stbi__err("bad AC huff","Corrupt JPEG"); + z->order[i] = which; + } + + { + int aa; + z->spec_start = stbi__get8(z->s); + z->spec_end = stbi__get8(z->s); // should be 63, but might be 0 + aa = stbi__get8(z->s); + z->succ_high = (aa >> 4); + z->succ_low = (aa & 15); + if (z->progressive) { + if (z->spec_start > 63 || z->spec_end > 63 || z->spec_start > z->spec_end || z->succ_high > 13 || z->succ_low > 13) + return stbi__err("bad SOS", "Corrupt JPEG"); + } else { + if (z->spec_start != 0) return stbi__err("bad SOS","Corrupt JPEG"); + if (z->succ_high != 0 || z->succ_low != 0) return stbi__err("bad SOS","Corrupt JPEG"); + z->spec_end = 63; + } + } + + return 1; +} + +static int stbi__free_jpeg_components(stbi__jpeg *z, int ncomp, int why) +{ + int i; + for (i=0; i < ncomp; ++i) { + if (z->img_comp[i].raw_data) { + STBI_FREE(z->img_comp[i].raw_data); + z->img_comp[i].raw_data = NULL; + z->img_comp[i].data = NULL; + } + if (z->img_comp[i].raw_coeff) { + STBI_FREE(z->img_comp[i].raw_coeff); + z->img_comp[i].raw_coeff = 0; + z->img_comp[i].coeff = 0; + } + if (z->img_comp[i].linebuf) { + STBI_FREE(z->img_comp[i].linebuf); + z->img_comp[i].linebuf = NULL; + } + } + return why; +} + +static int stbi__process_frame_header(stbi__jpeg *z, int scan) +{ + stbi__context *s = z->s; + int Lf,p,i,q, h_max=1,v_max=1,c; + Lf = stbi__get16be(s); if (Lf < 11) return stbi__err("bad SOF len","Corrupt JPEG"); // JPEG + p = stbi__get8(s); if (p != 8) return stbi__err("only 8-bit","JPEG format not supported: 8-bit only"); // JPEG baseline + s->img_y = stbi__get16be(s); if (s->img_y == 0) return stbi__err("no header height", "JPEG format not supported: delayed height"); // Legal, but we don't handle it--but neither does IJG + s->img_x = stbi__get16be(s); if (s->img_x == 0) return stbi__err("0 width","Corrupt JPEG"); // JPEG requires + if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)"); + c = stbi__get8(s); + if (c != 3 && c != 1 && c != 4) return stbi__err("bad component count","Corrupt JPEG"); + s->img_n = c; + for (i=0; i < c; ++i) { + z->img_comp[i].data = NULL; + z->img_comp[i].linebuf = NULL; + } + + if (Lf != 8+3*s->img_n) return stbi__err("bad SOF len","Corrupt JPEG"); + + z->rgb = 0; + for (i=0; i < s->img_n; ++i) { + static const unsigned char rgb[3] = { 'R', 'G', 'B' }; + z->img_comp[i].id = stbi__get8(s); + if (s->img_n == 3 && z->img_comp[i].id == rgb[i]) + ++z->rgb; + q = stbi__get8(s); + z->img_comp[i].h = (q >> 4); if (!z->img_comp[i].h || z->img_comp[i].h > 4) return stbi__err("bad H","Corrupt JPEG"); + z->img_comp[i].v = q & 15; if (!z->img_comp[i].v || z->img_comp[i].v > 4) return stbi__err("bad V","Corrupt JPEG"); + z->img_comp[i].tq = stbi__get8(s); if (z->img_comp[i].tq > 3) return stbi__err("bad TQ","Corrupt JPEG"); + } + + if (scan != STBI__SCAN_load) return 1; + + if (!stbi__mad3sizes_valid(s->img_x, s->img_y, s->img_n, 0)) return stbi__err("too large", "Image too large to decode"); + + for (i=0; i < s->img_n; ++i) { + if (z->img_comp[i].h > h_max) h_max = z->img_comp[i].h; + if (z->img_comp[i].v > v_max) v_max = z->img_comp[i].v; + } + + // check that plane subsampling factors are integer ratios; our resamplers can't deal with fractional ratios + // and I've never seen a non-corrupted JPEG file actually use them + for (i=0; i < s->img_n; ++i) { + if (h_max % z->img_comp[i].h != 0) return stbi__err("bad H","Corrupt JPEG"); + if (v_max % z->img_comp[i].v != 0) return stbi__err("bad V","Corrupt JPEG"); + } + + // compute interleaved mcu info + z->img_h_max = h_max; + z->img_v_max = v_max; + z->img_mcu_w = h_max * 8; + z->img_mcu_h = v_max * 8; + // these sizes can't be more than 17 bits + z->img_mcu_x = (s->img_x + z->img_mcu_w-1) / z->img_mcu_w; + z->img_mcu_y = (s->img_y + z->img_mcu_h-1) / z->img_mcu_h; + + for (i=0; i < s->img_n; ++i) { + // number of effective pixels (e.g. for non-interleaved MCU) + z->img_comp[i].x = (s->img_x * z->img_comp[i].h + h_max-1) / h_max; + z->img_comp[i].y = (s->img_y * z->img_comp[i].v + v_max-1) / v_max; + // to simplify generation, we'll allocate enough memory to decode + // the bogus oversized data from using interleaved MCUs and their + // big blocks (e.g. a 16x16 iMCU on an image of width 33); we won't + // discard the extra data until colorspace conversion + // + // img_mcu_x, img_mcu_y: <=17 bits; comp[i].h and .v are <=4 (checked earlier) + // so these muls can't overflow with 32-bit ints (which we require) + z->img_comp[i].w2 = z->img_mcu_x * z->img_comp[i].h * 8; + z->img_comp[i].h2 = z->img_mcu_y * z->img_comp[i].v * 8; + z->img_comp[i].coeff = 0; + z->img_comp[i].raw_coeff = 0; + z->img_comp[i].linebuf = NULL; + z->img_comp[i].raw_data = stbi__malloc_mad2(z->img_comp[i].w2, z->img_comp[i].h2, 15); + if (z->img_comp[i].raw_data == NULL) + return stbi__free_jpeg_components(z, i+1, stbi__err("outofmem", "Out of memory")); + // align blocks for idct using mmx/sse + z->img_comp[i].data = (stbi_uc*) (((size_t) z->img_comp[i].raw_data + 15) & ~15); + if (z->progressive) { + // w2, h2 are multiples of 8 (see above) + z->img_comp[i].coeff_w = z->img_comp[i].w2 / 8; + z->img_comp[i].coeff_h = z->img_comp[i].h2 / 8; + z->img_comp[i].raw_coeff = stbi__malloc_mad3(z->img_comp[i].w2, z->img_comp[i].h2, sizeof(short), 15); + if (z->img_comp[i].raw_coeff == NULL) + return stbi__free_jpeg_components(z, i+1, stbi__err("outofmem", "Out of memory")); + z->img_comp[i].coeff = (short*) (((size_t) z->img_comp[i].raw_coeff + 15) & ~15); + } + } + + return 1; +} + +// use comparisons since in some cases we handle more than one case (e.g. SOF) +#define stbi__DNL(x) ((x) == 0xdc) +#define stbi__SOI(x) ((x) == 0xd8) +#define stbi__EOI(x) ((x) == 0xd9) +#define stbi__SOF(x) ((x) == 0xc0 || (x) == 0xc1 || (x) == 0xc2) +#define stbi__SOS(x) ((x) == 0xda) + +#define stbi__SOF_progressive(x) ((x) == 0xc2) + +static int stbi__decode_jpeg_header(stbi__jpeg *z, int scan) +{ + int m; + z->jfif = 0; + z->app14_color_transform = -1; // valid values are 0,1,2 + z->marker = STBI__MARKER_none; // initialize cached marker to empty + m = stbi__get_marker(z); + if (!stbi__SOI(m)) return stbi__err("no SOI","Corrupt JPEG"); + if (scan == STBI__SCAN_type) return 1; + m = stbi__get_marker(z); + while (!stbi__SOF(m)) { + if (!stbi__process_marker(z,m)) return 0; + m = stbi__get_marker(z); + while (m == STBI__MARKER_none) { + // some files have extra padding after their blocks, so ok, we'll scan + if (stbi__at_eof(z->s)) return stbi__err("no SOF", "Corrupt JPEG"); + m = stbi__get_marker(z); + } + } + z->progressive = stbi__SOF_progressive(m); + if (!stbi__process_frame_header(z, scan)) return 0; + return 1; +} + +static stbi_uc stbi__skip_jpeg_junk_at_end(stbi__jpeg *j) +{ + // some JPEGs have junk at end, skip over it but if we find what looks + // like a valid marker, resume there + while (!stbi__at_eof(j->s)) { + stbi_uc x = stbi__get8(j->s); + while (x == 0xff) { // might be a marker + if (stbi__at_eof(j->s)) return STBI__MARKER_none; + x = stbi__get8(j->s); + if (x != 0x00 && x != 0xff) { + // not a stuffed zero or lead-in to another marker, looks + // like an actual marker, return it + return x; + } + // stuffed zero has x=0 now which ends the loop, meaning we go + // back to regular scan loop. + // repeated 0xff keeps trying to read the next byte of the marker. + } + } + return STBI__MARKER_none; +} + +// decode image to YCbCr format +static int stbi__decode_jpeg_image(stbi__jpeg *j) +{ + int m; + for (m = 0; m < 4; m++) { + j->img_comp[m].raw_data = NULL; + j->img_comp[m].raw_coeff = NULL; + } + j->restart_interval = 0; + if (!stbi__decode_jpeg_header(j, STBI__SCAN_load)) return 0; + m = stbi__get_marker(j); + while (!stbi__EOI(m)) { + if (stbi__SOS(m)) { + if (!stbi__process_scan_header(j)) return 0; + if (!stbi__parse_entropy_coded_data(j)) return 0; + if (j->marker == STBI__MARKER_none ) { + j->marker = stbi__skip_jpeg_junk_at_end(j); + // if we reach eof without hitting a marker, stbi__get_marker() below will fail and we'll eventually return 0 + } + m = stbi__get_marker(j); + if (STBI__RESTART(m)) + m = stbi__get_marker(j); + } else if (stbi__DNL(m)) { + int Ld = stbi__get16be(j->s); + stbi__uint32 NL = stbi__get16be(j->s); + if (Ld != 4) return stbi__err("bad DNL len", "Corrupt JPEG"); + if (NL != j->s->img_y) return stbi__err("bad DNL height", "Corrupt JPEG"); + m = stbi__get_marker(j); + } else { + if (!stbi__process_marker(j, m)) return 1; + m = stbi__get_marker(j); + } + } + if (j->progressive) + stbi__jpeg_finish(j); + return 1; +} + +// static jfif-centered resampling (across block boundaries) + +typedef stbi_uc *(*resample_row_func)(stbi_uc *out, stbi_uc *in0, stbi_uc *in1, + int w, int hs); + +#define stbi__div4(x) ((stbi_uc) ((x) >> 2)) + +static stbi_uc *resample_row_1(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) +{ + STBI_NOTUSED(out); + STBI_NOTUSED(in_far); + STBI_NOTUSED(w); + STBI_NOTUSED(hs); + return in_near; +} + +static stbi_uc* stbi__resample_row_v_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) +{ + // need to generate two samples vertically for every one in input + int i; + STBI_NOTUSED(hs); + for (i=0; i < w; ++i) + out[i] = stbi__div4(3*in_near[i] + in_far[i] + 2); + return out; +} + +static stbi_uc* stbi__resample_row_h_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) +{ + // need to generate two samples horizontally for every one in input + int i; + stbi_uc *input = in_near; + + if (w == 1) { + // if only one sample, can't do any interpolation + out[0] = out[1] = input[0]; + return out; + } + + out[0] = input[0]; + out[1] = stbi__div4(input[0]*3 + input[1] + 2); + for (i=1; i < w-1; ++i) { + int n = 3*input[i]+2; + out[i*2+0] = stbi__div4(n+input[i-1]); + out[i*2+1] = stbi__div4(n+input[i+1]); + } + out[i*2+0] = stbi__div4(input[w-2]*3 + input[w-1] + 2); + out[i*2+1] = input[w-1]; + + STBI_NOTUSED(in_far); + STBI_NOTUSED(hs); + + return out; +} + +#define stbi__div16(x) ((stbi_uc) ((x) >> 4)) + +static stbi_uc *stbi__resample_row_hv_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) +{ + // need to generate 2x2 samples for every one in input + int i,t0,t1; + if (w == 1) { + out[0] = out[1] = stbi__div4(3*in_near[0] + in_far[0] + 2); + return out; + } + + t1 = 3*in_near[0] + in_far[0]; + out[0] = stbi__div4(t1+2); + for (i=1; i < w; ++i) { + t0 = t1; + t1 = 3*in_near[i]+in_far[i]; + out[i*2-1] = stbi__div16(3*t0 + t1 + 8); + out[i*2 ] = stbi__div16(3*t1 + t0 + 8); + } + out[w*2-1] = stbi__div4(t1+2); + + STBI_NOTUSED(hs); + + return out; +} + +#if defined(STBI_SSE2) || defined(STBI_NEON) +static stbi_uc *stbi__resample_row_hv_2_simd(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) +{ + // need to generate 2x2 samples for every one in input + int i=0,t0,t1; + + if (w == 1) { + out[0] = out[1] = stbi__div4(3*in_near[0] + in_far[0] + 2); + return out; + } + + t1 = 3*in_near[0] + in_far[0]; + // process groups of 8 pixels for as long as we can. + // note we can't handle the last pixel in a row in this loop + // because we need to handle the filter boundary conditions. + for (; i < ((w-1) & ~7); i += 8) { +#if defined(STBI_SSE2) + // load and perform the vertical filtering pass + // this uses 3*x + y = 4*x + (y - x) + __m128i zero = _mm_setzero_si128(); + __m128i farb = _mm_loadl_epi64((__m128i *) (in_far + i)); + __m128i nearb = _mm_loadl_epi64((__m128i *) (in_near + i)); + __m128i farw = _mm_unpacklo_epi8(farb, zero); + __m128i nearw = _mm_unpacklo_epi8(nearb, zero); + __m128i diff = _mm_sub_epi16(farw, nearw); + __m128i nears = _mm_slli_epi16(nearw, 2); + __m128i curr = _mm_add_epi16(nears, diff); // current row + + // horizontal filter works the same based on shifted vers of current + // row. "prev" is current row shifted right by 1 pixel; we need to + // insert the previous pixel value (from t1). + // "next" is current row shifted left by 1 pixel, with first pixel + // of next block of 8 pixels added in. + __m128i prv0 = _mm_slli_si128(curr, 2); + __m128i nxt0 = _mm_srli_si128(curr, 2); + __m128i prev = _mm_insert_epi16(prv0, t1, 0); + __m128i next = _mm_insert_epi16(nxt0, 3*in_near[i+8] + in_far[i+8], 7); + + // horizontal filter, polyphase implementation since it's convenient: + // even pixels = 3*cur + prev = cur*4 + (prev - cur) + // odd pixels = 3*cur + next = cur*4 + (next - cur) + // note the shared term. + __m128i bias = _mm_set1_epi16(8); + __m128i curs = _mm_slli_epi16(curr, 2); + __m128i prvd = _mm_sub_epi16(prev, curr); + __m128i nxtd = _mm_sub_epi16(next, curr); + __m128i curb = _mm_add_epi16(curs, bias); + __m128i even = _mm_add_epi16(prvd, curb); + __m128i odd = _mm_add_epi16(nxtd, curb); + + // interleave even and odd pixels, then undo scaling. + __m128i int0 = _mm_unpacklo_epi16(even, odd); + __m128i int1 = _mm_unpackhi_epi16(even, odd); + __m128i de0 = _mm_srli_epi16(int0, 4); + __m128i de1 = _mm_srli_epi16(int1, 4); + + // pack and write output + __m128i outv = _mm_packus_epi16(de0, de1); + _mm_storeu_si128((__m128i *) (out + i*2), outv); +#elif defined(STBI_NEON) + // load and perform the vertical filtering pass + // this uses 3*x + y = 4*x + (y - x) + uint8x8_t farb = vld1_u8(in_far + i); + uint8x8_t nearb = vld1_u8(in_near + i); + int16x8_t diff = vreinterpretq_s16_u16(vsubl_u8(farb, nearb)); + int16x8_t nears = vreinterpretq_s16_u16(vshll_n_u8(nearb, 2)); + int16x8_t curr = vaddq_s16(nears, diff); // current row + + // horizontal filter works the same based on shifted vers of current + // row. "prev" is current row shifted right by 1 pixel; we need to + // insert the previous pixel value (from t1). + // "next" is current row shifted left by 1 pixel, with first pixel + // of next block of 8 pixels added in. + int16x8_t prv0 = vextq_s16(curr, curr, 7); + int16x8_t nxt0 = vextq_s16(curr, curr, 1); + int16x8_t prev = vsetq_lane_s16(t1, prv0, 0); + int16x8_t next = vsetq_lane_s16(3*in_near[i+8] + in_far[i+8], nxt0, 7); + + // horizontal filter, polyphase implementation since it's convenient: + // even pixels = 3*cur + prev = cur*4 + (prev - cur) + // odd pixels = 3*cur + next = cur*4 + (next - cur) + // note the shared term. + int16x8_t curs = vshlq_n_s16(curr, 2); + int16x8_t prvd = vsubq_s16(prev, curr); + int16x8_t nxtd = vsubq_s16(next, curr); + int16x8_t even = vaddq_s16(curs, prvd); + int16x8_t odd = vaddq_s16(curs, nxtd); + + // undo scaling and round, then store with even/odd phases interleaved + uint8x8x2_t o; + o.val[0] = vqrshrun_n_s16(even, 4); + o.val[1] = vqrshrun_n_s16(odd, 4); + vst2_u8(out + i*2, o); +#endif + + // "previous" value for next iter + t1 = 3*in_near[i+7] + in_far[i+7]; + } + + t0 = t1; + t1 = 3*in_near[i] + in_far[i]; + out[i*2] = stbi__div16(3*t1 + t0 + 8); + + for (++i; i < w; ++i) { + t0 = t1; + t1 = 3*in_near[i]+in_far[i]; + out[i*2-1] = stbi__div16(3*t0 + t1 + 8); + out[i*2 ] = stbi__div16(3*t1 + t0 + 8); + } + out[w*2-1] = stbi__div4(t1+2); + + STBI_NOTUSED(hs); + + return out; +} +#endif + +static stbi_uc *stbi__resample_row_generic(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) +{ + // resample with nearest-neighbor + int i,j; + STBI_NOTUSED(in_far); + for (i=0; i < w; ++i) + for (j=0; j < hs; ++j) + out[i*hs+j] = in_near[i]; + return out; +} + +// this is a reduced-precision calculation of YCbCr-to-RGB introduced +// to make sure the code produces the same results in both SIMD and scalar +#define stbi__float2fixed(x) (((int) ((x) * 4096.0f + 0.5f)) << 8) +static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step) +{ + int i; + for (i=0; i < count; ++i) { + int y_fixed = (y[i] << 20) + (1<<19); // rounding + int r,g,b; + int cr = pcr[i] - 128; + int cb = pcb[i] - 128; + r = y_fixed + cr* stbi__float2fixed(1.40200f); + g = y_fixed + (cr*-stbi__float2fixed(0.71414f)) + ((cb*-stbi__float2fixed(0.34414f)) & 0xffff0000); + b = y_fixed + cb* stbi__float2fixed(1.77200f); + r >>= 20; + g >>= 20; + b >>= 20; + if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; } + if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; } + if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; } + out[0] = (stbi_uc)r; + out[1] = (stbi_uc)g; + out[2] = (stbi_uc)b; + out[3] = 255; + out += step; + } +} + +#if defined(STBI_SSE2) || defined(STBI_NEON) +static void stbi__YCbCr_to_RGB_simd(stbi_uc *out, stbi_uc const *y, stbi_uc const *pcb, stbi_uc const *pcr, int count, int step) +{ + int i = 0; + +#ifdef STBI_SSE2 + // step == 3 is pretty ugly on the final interleave, and i'm not convinced + // it's useful in practice (you wouldn't use it for textures, for example). + // so just accelerate step == 4 case. + if (step == 4) { + // this is a fairly straightforward implementation and not super-optimized. + __m128i signflip = _mm_set1_epi8(-0x80); + __m128i cr_const0 = _mm_set1_epi16( (short) ( 1.40200f*4096.0f+0.5f)); + __m128i cr_const1 = _mm_set1_epi16( - (short) ( 0.71414f*4096.0f+0.5f)); + __m128i cb_const0 = _mm_set1_epi16( - (short) ( 0.34414f*4096.0f+0.5f)); + __m128i cb_const1 = _mm_set1_epi16( (short) ( 1.77200f*4096.0f+0.5f)); + __m128i y_bias = _mm_set1_epi8((char) (unsigned char) 128); + __m128i xw = _mm_set1_epi16(255); // alpha channel + + for (; i+7 < count; i += 8) { + // load + __m128i y_bytes = _mm_loadl_epi64((__m128i *) (y+i)); + __m128i cr_bytes = _mm_loadl_epi64((__m128i *) (pcr+i)); + __m128i cb_bytes = _mm_loadl_epi64((__m128i *) (pcb+i)); + __m128i cr_biased = _mm_xor_si128(cr_bytes, signflip); // -128 + __m128i cb_biased = _mm_xor_si128(cb_bytes, signflip); // -128 + + // unpack to short (and left-shift cr, cb by 8) + __m128i yw = _mm_unpacklo_epi8(y_bias, y_bytes); + __m128i crw = _mm_unpacklo_epi8(_mm_setzero_si128(), cr_biased); + __m128i cbw = _mm_unpacklo_epi8(_mm_setzero_si128(), cb_biased); + + // color transform + __m128i yws = _mm_srli_epi16(yw, 4); + __m128i cr0 = _mm_mulhi_epi16(cr_const0, crw); + __m128i cb0 = _mm_mulhi_epi16(cb_const0, cbw); + __m128i cb1 = _mm_mulhi_epi16(cbw, cb_const1); + __m128i cr1 = _mm_mulhi_epi16(crw, cr_const1); + __m128i rws = _mm_add_epi16(cr0, yws); + __m128i gwt = _mm_add_epi16(cb0, yws); + __m128i bws = _mm_add_epi16(yws, cb1); + __m128i gws = _mm_add_epi16(gwt, cr1); + + // descale + __m128i rw = _mm_srai_epi16(rws, 4); + __m128i bw = _mm_srai_epi16(bws, 4); + __m128i gw = _mm_srai_epi16(gws, 4); + + // back to byte, set up for transpose + __m128i brb = _mm_packus_epi16(rw, bw); + __m128i gxb = _mm_packus_epi16(gw, xw); + + // transpose to interleave channels + __m128i t0 = _mm_unpacklo_epi8(brb, gxb); + __m128i t1 = _mm_unpackhi_epi8(brb, gxb); + __m128i o0 = _mm_unpacklo_epi16(t0, t1); + __m128i o1 = _mm_unpackhi_epi16(t0, t1); + + // store + _mm_storeu_si128((__m128i *) (out + 0), o0); + _mm_storeu_si128((__m128i *) (out + 16), o1); + out += 32; + } + } +#endif + +#ifdef STBI_NEON + // in this version, step=3 support would be easy to add. but is there demand? + if (step == 4) { + // this is a fairly straightforward implementation and not super-optimized. + uint8x8_t signflip = vdup_n_u8(0x80); + int16x8_t cr_const0 = vdupq_n_s16( (short) ( 1.40200f*4096.0f+0.5f)); + int16x8_t cr_const1 = vdupq_n_s16( - (short) ( 0.71414f*4096.0f+0.5f)); + int16x8_t cb_const0 = vdupq_n_s16( - (short) ( 0.34414f*4096.0f+0.5f)); + int16x8_t cb_const1 = vdupq_n_s16( (short) ( 1.77200f*4096.0f+0.5f)); + + for (; i+7 < count; i += 8) { + // load + uint8x8_t y_bytes = vld1_u8(y + i); + uint8x8_t cr_bytes = vld1_u8(pcr + i); + uint8x8_t cb_bytes = vld1_u8(pcb + i); + int8x8_t cr_biased = vreinterpret_s8_u8(vsub_u8(cr_bytes, signflip)); + int8x8_t cb_biased = vreinterpret_s8_u8(vsub_u8(cb_bytes, signflip)); + + // expand to s16 + int16x8_t yws = vreinterpretq_s16_u16(vshll_n_u8(y_bytes, 4)); + int16x8_t crw = vshll_n_s8(cr_biased, 7); + int16x8_t cbw = vshll_n_s8(cb_biased, 7); + + // color transform + int16x8_t cr0 = vqdmulhq_s16(crw, cr_const0); + int16x8_t cb0 = vqdmulhq_s16(cbw, cb_const0); + int16x8_t cr1 = vqdmulhq_s16(crw, cr_const1); + int16x8_t cb1 = vqdmulhq_s16(cbw, cb_const1); + int16x8_t rws = vaddq_s16(yws, cr0); + int16x8_t gws = vaddq_s16(vaddq_s16(yws, cb0), cr1); + int16x8_t bws = vaddq_s16(yws, cb1); + + // undo scaling, round, convert to byte + uint8x8x4_t o; + o.val[0] = vqrshrun_n_s16(rws, 4); + o.val[1] = vqrshrun_n_s16(gws, 4); + o.val[2] = vqrshrun_n_s16(bws, 4); + o.val[3] = vdup_n_u8(255); + + // store, interleaving r/g/b/a + vst4_u8(out, o); + out += 8*4; + } + } +#endif + + for (; i < count; ++i) { + int y_fixed = (y[i] << 20) + (1<<19); // rounding + int r,g,b; + int cr = pcr[i] - 128; + int cb = pcb[i] - 128; + r = y_fixed + cr* stbi__float2fixed(1.40200f); + g = y_fixed + cr*-stbi__float2fixed(0.71414f) + ((cb*-stbi__float2fixed(0.34414f)) & 0xffff0000); + b = y_fixed + cb* stbi__float2fixed(1.77200f); + r >>= 20; + g >>= 20; + b >>= 20; + if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; } + if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; } + if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; } + out[0] = (stbi_uc)r; + out[1] = (stbi_uc)g; + out[2] = (stbi_uc)b; + out[3] = 255; + out += step; + } +} +#endif + +// set up the kernels +static void stbi__setup_jpeg(stbi__jpeg *j) +{ + j->idct_block_kernel = stbi__idct_block; + j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_row; + j->resample_row_hv_2_kernel = stbi__resample_row_hv_2; + +#ifdef STBI_SSE2 + if (stbi__sse2_available()) { + j->idct_block_kernel = stbi__idct_simd; + j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; + j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; + } +#endif + +#ifdef STBI_NEON + j->idct_block_kernel = stbi__idct_simd; + j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; + j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; +#endif +} + +// clean up the temporary component buffers +static void stbi__cleanup_jpeg(stbi__jpeg *j) +{ + stbi__free_jpeg_components(j, j->s->img_n, 0); +} + +typedef struct +{ + resample_row_func resample; + stbi_uc *line0,*line1; + int hs,vs; // expansion factor in each axis + int w_lores; // horizontal pixels pre-expansion + int ystep; // how far through vertical expansion we are + int ypos; // which pre-expansion row we're on +} stbi__resample; + +// fast 0..255 * 0..255 => 0..255 rounded multiplication +static stbi_uc stbi__blinn_8x8(stbi_uc x, stbi_uc y) +{ + unsigned int t = x*y + 128; + return (stbi_uc) ((t + (t >>8)) >> 8); +} + +static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp, int req_comp) +{ + int n, decode_n, is_rgb; + z->s->img_n = 0; // make stbi__cleanup_jpeg safe + + // validate req_comp + if (req_comp < 0 || req_comp > 4) return stbi__errpuc("bad req_comp", "Internal error"); + + // load a jpeg image from whichever source, but leave in YCbCr format + if (!stbi__decode_jpeg_image(z)) { stbi__cleanup_jpeg(z); return NULL; } + + // determine actual number of components to generate + n = req_comp ? req_comp : z->s->img_n >= 3 ? 3 : 1; + + is_rgb = z->s->img_n == 3 && (z->rgb == 3 || (z->app14_color_transform == 0 && !z->jfif)); + + if (z->s->img_n == 3 && n < 3 && !is_rgb) + decode_n = 1; + else + decode_n = z->s->img_n; + + // nothing to do if no components requested; check this now to avoid + // accessing uninitialized coutput[0] later + if (decode_n <= 0) { stbi__cleanup_jpeg(z); return NULL; } + + // resample and color-convert + { + int k; + unsigned int i,j; + stbi_uc *output; + stbi_uc *coutput[4] = { NULL, NULL, NULL, NULL }; + + stbi__resample res_comp[4]; + + for (k=0; k < decode_n; ++k) { + stbi__resample *r = &res_comp[k]; + + // allocate line buffer big enough for upsampling off the edges + // with upsample factor of 4 + z->img_comp[k].linebuf = (stbi_uc *) stbi__malloc(z->s->img_x + 3); + if (!z->img_comp[k].linebuf) { stbi__cleanup_jpeg(z); return stbi__errpuc("outofmem", "Out of memory"); } + + r->hs = z->img_h_max / z->img_comp[k].h; + r->vs = z->img_v_max / z->img_comp[k].v; + r->ystep = r->vs >> 1; + r->w_lores = (z->s->img_x + r->hs-1) / r->hs; + r->ypos = 0; + r->line0 = r->line1 = z->img_comp[k].data; + + if (r->hs == 1 && r->vs == 1) r->resample = resample_row_1; + else if (r->hs == 1 && r->vs == 2) r->resample = stbi__resample_row_v_2; + else if (r->hs == 2 && r->vs == 1) r->resample = stbi__resample_row_h_2; + else if (r->hs == 2 && r->vs == 2) r->resample = z->resample_row_hv_2_kernel; + else r->resample = stbi__resample_row_generic; + } + + // can't error after this so, this is safe + output = (stbi_uc *) stbi__malloc_mad3(n, z->s->img_x, z->s->img_y, 1); + if (!output) { stbi__cleanup_jpeg(z); return stbi__errpuc("outofmem", "Out of memory"); } + + // now go ahead and resample + for (j=0; j < z->s->img_y; ++j) { + stbi_uc *out = output + n * z->s->img_x * j; + for (k=0; k < decode_n; ++k) { + stbi__resample *r = &res_comp[k]; + int y_bot = r->ystep >= (r->vs >> 1); + coutput[k] = r->resample(z->img_comp[k].linebuf, + y_bot ? r->line1 : r->line0, + y_bot ? r->line0 : r->line1, + r->w_lores, r->hs); + if (++r->ystep >= r->vs) { + r->ystep = 0; + r->line0 = r->line1; + if (++r->ypos < z->img_comp[k].y) + r->line1 += z->img_comp[k].w2; + } + } + if (n >= 3) { + stbi_uc *y = coutput[0]; + if (z->s->img_n == 3) { + if (is_rgb) { + for (i=0; i < z->s->img_x; ++i) { + out[0] = y[i]; + out[1] = coutput[1][i]; + out[2] = coutput[2][i]; + out[3] = 255; + out += n; + } + } else { + z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); + } + } else if (z->s->img_n == 4) { + if (z->app14_color_transform == 0) { // CMYK + for (i=0; i < z->s->img_x; ++i) { + stbi_uc m = coutput[3][i]; + out[0] = stbi__blinn_8x8(coutput[0][i], m); + out[1] = stbi__blinn_8x8(coutput[1][i], m); + out[2] = stbi__blinn_8x8(coutput[2][i], m); + out[3] = 255; + out += n; + } + } else if (z->app14_color_transform == 2) { // YCCK + z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); + for (i=0; i < z->s->img_x; ++i) { + stbi_uc m = coutput[3][i]; + out[0] = stbi__blinn_8x8(255 - out[0], m); + out[1] = stbi__blinn_8x8(255 - out[1], m); + out[2] = stbi__blinn_8x8(255 - out[2], m); + out += n; + } + } else { // YCbCr + alpha? Ignore the fourth channel for now + z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); + } + } else + for (i=0; i < z->s->img_x; ++i) { + out[0] = out[1] = out[2] = y[i]; + out[3] = 255; // not used if n==3 + out += n; + } + } else { + if (is_rgb) { + if (n == 1) + for (i=0; i < z->s->img_x; ++i) + *out++ = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]); + else { + for (i=0; i < z->s->img_x; ++i, out += 2) { + out[0] = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]); + out[1] = 255; + } + } + } else if (z->s->img_n == 4 && z->app14_color_transform == 0) { + for (i=0; i < z->s->img_x; ++i) { + stbi_uc m = coutput[3][i]; + stbi_uc r = stbi__blinn_8x8(coutput[0][i], m); + stbi_uc g = stbi__blinn_8x8(coutput[1][i], m); + stbi_uc b = stbi__blinn_8x8(coutput[2][i], m); + out[0] = stbi__compute_y(r, g, b); + out[1] = 255; + out += n; + } + } else if (z->s->img_n == 4 && z->app14_color_transform == 2) { + for (i=0; i < z->s->img_x; ++i) { + out[0] = stbi__blinn_8x8(255 - coutput[0][i], coutput[3][i]); + out[1] = 255; + out += n; + } + } else { + stbi_uc *y = coutput[0]; + if (n == 1) + for (i=0; i < z->s->img_x; ++i) out[i] = y[i]; + else + for (i=0; i < z->s->img_x; ++i) { *out++ = y[i]; *out++ = 255; } + } + } + } + stbi__cleanup_jpeg(z); + *out_x = z->s->img_x; + *out_y = z->s->img_y; + if (comp) *comp = z->s->img_n >= 3 ? 3 : 1; // report original components, not output + return output; + } +} + +static void *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri) +{ + unsigned char* result; + stbi__jpeg* j = (stbi__jpeg*) stbi__malloc(sizeof(stbi__jpeg)); + if (!j) return stbi__errpuc("outofmem", "Out of memory"); + memset(j, 0, sizeof(stbi__jpeg)); + STBI_NOTUSED(ri); + j->s = s; + stbi__setup_jpeg(j); + result = load_jpeg_image(j, x,y,comp,req_comp); + STBI_FREE(j); + return result; +} + +static int stbi__jpeg_test(stbi__context *s) +{ + int r; + stbi__jpeg* j = (stbi__jpeg*)stbi__malloc(sizeof(stbi__jpeg)); + if (!j) return stbi__err("outofmem", "Out of memory"); + memset(j, 0, sizeof(stbi__jpeg)); + j->s = s; + stbi__setup_jpeg(j); + r = stbi__decode_jpeg_header(j, STBI__SCAN_type); + stbi__rewind(s); + STBI_FREE(j); + return r; +} + +static int stbi__jpeg_info_raw(stbi__jpeg *j, int *x, int *y, int *comp) +{ + if (!stbi__decode_jpeg_header(j, STBI__SCAN_header)) { + stbi__rewind( j->s ); + return 0; + } + if (x) *x = j->s->img_x; + if (y) *y = j->s->img_y; + if (comp) *comp = j->s->img_n >= 3 ? 3 : 1; + return 1; +} + +static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp) +{ + int result; + stbi__jpeg* j = (stbi__jpeg*) (stbi__malloc(sizeof(stbi__jpeg))); + if (!j) return stbi__err("outofmem", "Out of memory"); + memset(j, 0, sizeof(stbi__jpeg)); + j->s = s; + result = stbi__jpeg_info_raw(j, x, y, comp); + STBI_FREE(j); + return result; +} +#endif + +// public domain zlib decode v0.2 Sean Barrett 2006-11-18 +// simple implementation +// - all input must be provided in an upfront buffer +// - all output is written to a single output buffer (can malloc/realloc) +// performance +// - fast huffman + +#ifndef STBI_NO_ZLIB + +// fast-way is faster to check than jpeg huffman, but slow way is slower +#define STBI__ZFAST_BITS 9 // accelerate all cases in default tables +#define STBI__ZFAST_MASK ((1 << STBI__ZFAST_BITS) - 1) +#define STBI__ZNSYMS 288 // number of symbols in literal/length alphabet + +// zlib-style huffman encoding +// (jpegs packs from left, zlib from right, so can't share code) +typedef struct +{ + stbi__uint16 fast[1 << STBI__ZFAST_BITS]; + stbi__uint16 firstcode[16]; + int maxcode[17]; + stbi__uint16 firstsymbol[16]; + stbi_uc size[STBI__ZNSYMS]; + stbi__uint16 value[STBI__ZNSYMS]; +} stbi__zhuffman; + +stbi_inline static int stbi__bitreverse16(int n) +{ + n = ((n & 0xAAAA) >> 1) | ((n & 0x5555) << 1); + n = ((n & 0xCCCC) >> 2) | ((n & 0x3333) << 2); + n = ((n & 0xF0F0) >> 4) | ((n & 0x0F0F) << 4); + n = ((n & 0xFF00) >> 8) | ((n & 0x00FF) << 8); + return n; +} + +stbi_inline static int stbi__bit_reverse(int v, int bits) +{ + STBI_ASSERT(bits <= 16); + // to bit reverse n bits, reverse 16 and shift + // e.g. 11 bits, bit reverse and shift away 5 + return stbi__bitreverse16(v) >> (16-bits); +} + +static int stbi__zbuild_huffman(stbi__zhuffman *z, const stbi_uc *sizelist, int num) +{ + int i,k=0; + int code, next_code[16], sizes[17]; + + // DEFLATE spec for generating codes + memset(sizes, 0, sizeof(sizes)); + memset(z->fast, 0, sizeof(z->fast)); + for (i=0; i < num; ++i) + ++sizes[sizelist[i]]; + sizes[0] = 0; + for (i=1; i < 16; ++i) + if (sizes[i] > (1 << i)) + return stbi__err("bad sizes", "Corrupt PNG"); + code = 0; + for (i=1; i < 16; ++i) { + next_code[i] = code; + z->firstcode[i] = (stbi__uint16) code; + z->firstsymbol[i] = (stbi__uint16) k; + code = (code + sizes[i]); + if (sizes[i]) + if (code-1 >= (1 << i)) return stbi__err("bad codelengths","Corrupt PNG"); + z->maxcode[i] = code << (16-i); // preshift for inner loop + code <<= 1; + k += sizes[i]; + } + z->maxcode[16] = 0x10000; // sentinel + for (i=0; i < num; ++i) { + int s = sizelist[i]; + if (s) { + int c = next_code[s] - z->firstcode[s] + z->firstsymbol[s]; + stbi__uint16 fastv = (stbi__uint16) ((s << 9) | i); + z->size [c] = (stbi_uc ) s; + z->value[c] = (stbi__uint16) i; + if (s <= STBI__ZFAST_BITS) { + int j = stbi__bit_reverse(next_code[s],s); + while (j < (1 << STBI__ZFAST_BITS)) { + z->fast[j] = fastv; + j += (1 << s); + } + } + ++next_code[s]; + } + } + return 1; +} + +// zlib-from-memory implementation for PNG reading +// because PNG allows splitting the zlib stream arbitrarily, +// and it's annoying structurally to have PNG call ZLIB call PNG, +// we require PNG read all the IDATs and combine them into a single +// memory buffer + +typedef struct +{ + stbi_uc *zbuffer, *zbuffer_end; + int num_bits; + int hit_zeof_once; + stbi__uint32 code_buffer; + + char *zout; + char *zout_start; + char *zout_end; + int z_expandable; + + stbi__zhuffman z_length, z_distance; +} stbi__zbuf; + +stbi_inline static int stbi__zeof(stbi__zbuf *z) +{ + return (z->zbuffer >= z->zbuffer_end); +} + +stbi_inline static stbi_uc stbi__zget8(stbi__zbuf *z) +{ + return stbi__zeof(z) ? 0 : *z->zbuffer++; +} + +static void stbi__fill_bits(stbi__zbuf *z) +{ + do { + if (z->code_buffer >= (1U << z->num_bits)) { + z->zbuffer = z->zbuffer_end; /* treat this as EOF so we fail. */ + return; + } + z->code_buffer |= (unsigned int) stbi__zget8(z) << z->num_bits; + z->num_bits += 8; + } while (z->num_bits <= 24); +} + +stbi_inline static unsigned int stbi__zreceive(stbi__zbuf *z, int n) +{ + unsigned int k; + if (z->num_bits < n) stbi__fill_bits(z); + k = z->code_buffer & ((1 << n) - 1); + z->code_buffer >>= n; + z->num_bits -= n; + return k; +} + +static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z) +{ + int b,s,k; + // not resolved by fast table, so compute it the slow way + // use jpeg approach, which requires MSbits at top + k = stbi__bit_reverse(a->code_buffer, 16); + for (s=STBI__ZFAST_BITS+1; ; ++s) + if (k < z->maxcode[s]) + break; + if (s >= 16) return -1; // invalid code! + // code size is s, so: + b = (k >> (16-s)) - z->firstcode[s] + z->firstsymbol[s]; + if (b >= STBI__ZNSYMS) return -1; // some data was corrupt somewhere! + if (z->size[b] != s) return -1; // was originally an assert, but report failure instead. + a->code_buffer >>= s; + a->num_bits -= s; + return z->value[b]; +} + +stbi_inline static int stbi__zhuffman_decode(stbi__zbuf *a, stbi__zhuffman *z) +{ + int b,s; + if (a->num_bits < 16) { + if (stbi__zeof(a)) { + if (!a->hit_zeof_once) { + // This is the first time we hit eof, insert 16 extra padding btis + // to allow us to keep going; if we actually consume any of them + // though, that is invalid data. This is caught later. + a->hit_zeof_once = 1; + a->num_bits += 16; // add 16 implicit zero bits + } else { + // We already inserted our extra 16 padding bits and are again + // out, this stream is actually prematurely terminated. + return -1; + } + } else { + stbi__fill_bits(a); + } + } + b = z->fast[a->code_buffer & STBI__ZFAST_MASK]; + if (b) { + s = b >> 9; + a->code_buffer >>= s; + a->num_bits -= s; + return b & 511; + } + return stbi__zhuffman_decode_slowpath(a, z); +} + +static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) // need to make room for n bytes +{ + char *q; + unsigned int cur, limit, old_limit; + z->zout = zout; + if (!z->z_expandable) return stbi__err("output buffer limit","Corrupt PNG"); + cur = (unsigned int) (z->zout - z->zout_start); + limit = old_limit = (unsigned) (z->zout_end - z->zout_start); + if (UINT_MAX - cur < (unsigned) n) return stbi__err("outofmem", "Out of memory"); + while (cur + n > limit) { + if(limit > UINT_MAX / 2) return stbi__err("outofmem", "Out of memory"); + limit *= 2; + } + q = (char *) STBI_REALLOC_SIZED(z->zout_start, old_limit, limit); + STBI_NOTUSED(old_limit); + if (q == NULL) return stbi__err("outofmem", "Out of memory"); + z->zout_start = q; + z->zout = q + cur; + z->zout_end = q + limit; + return 1; +} + +static const int stbi__zlength_base[31] = { + 3,4,5,6,7,8,9,10,11,13, + 15,17,19,23,27,31,35,43,51,59, + 67,83,99,115,131,163,195,227,258,0,0 }; + +static const int stbi__zlength_extra[31]= +{ 0,0,0,0,0,0,0,0,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,5,5,5,5,0,0,0 }; + +static const int stbi__zdist_base[32] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193, +257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577,0,0}; + +static const int stbi__zdist_extra[32] = +{ 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13}; + +static int stbi__parse_huffman_block(stbi__zbuf *a) +{ + char *zout = a->zout; + for(;;) { + int z = stbi__zhuffman_decode(a, &a->z_length); + if (z < 256) { + if (z < 0) return stbi__err("bad huffman code","Corrupt PNG"); // error in huffman codes + if (zout >= a->zout_end) { + if (!stbi__zexpand(a, zout, 1)) return 0; + zout = a->zout; + } + *zout++ = (char) z; + } else { + stbi_uc *p; + int len,dist; + if (z == 256) { + a->zout = zout; + if (a->hit_zeof_once && a->num_bits < 16) { + // The first time we hit zeof, we inserted 16 extra zero bits into our bit + // buffer so the decoder can just do its speculative decoding. But if we + // actually consumed any of those bits (which is the case when num_bits < 16), + // the stream actually read past the end so it is malformed. + return stbi__err("unexpected end","Corrupt PNG"); + } + return 1; + } + if (z >= 286) return stbi__err("bad huffman code","Corrupt PNG"); // per DEFLATE, length codes 286 and 287 must not appear in compressed data + z -= 257; + len = stbi__zlength_base[z]; + if (stbi__zlength_extra[z]) len += stbi__zreceive(a, stbi__zlength_extra[z]); + z = stbi__zhuffman_decode(a, &a->z_distance); + if (z < 0 || z >= 30) return stbi__err("bad huffman code","Corrupt PNG"); // per DEFLATE, distance codes 30 and 31 must not appear in compressed data + dist = stbi__zdist_base[z]; + if (stbi__zdist_extra[z]) dist += stbi__zreceive(a, stbi__zdist_extra[z]); + if (zout - a->zout_start < dist) return stbi__err("bad dist","Corrupt PNG"); + if (len > a->zout_end - zout) { + if (!stbi__zexpand(a, zout, len)) return 0; + zout = a->zout; + } + p = (stbi_uc *) (zout - dist); + if (dist == 1) { // run of one byte; common in images. + stbi_uc v = *p; + if (len) { do *zout++ = v; while (--len); } + } else { + if (len) { do *zout++ = *p++; while (--len); } + } + } + } +} + +static int stbi__compute_huffman_codes(stbi__zbuf *a) +{ + static const stbi_uc length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 }; + stbi__zhuffman z_codelength; + stbi_uc lencodes[286+32+137];//padding for maximum single op + stbi_uc codelength_sizes[19]; + int i,n; + + int hlit = stbi__zreceive(a,5) + 257; + int hdist = stbi__zreceive(a,5) + 1; + int hclen = stbi__zreceive(a,4) + 4; + int ntot = hlit + hdist; + + memset(codelength_sizes, 0, sizeof(codelength_sizes)); + for (i=0; i < hclen; ++i) { + int s = stbi__zreceive(a,3); + codelength_sizes[length_dezigzag[i]] = (stbi_uc) s; + } + if (!stbi__zbuild_huffman(&z_codelength, codelength_sizes, 19)) return 0; + + n = 0; + while (n < ntot) { + int c = stbi__zhuffman_decode(a, &z_codelength); + if (c < 0 || c >= 19) return stbi__err("bad codelengths", "Corrupt PNG"); + if (c < 16) + lencodes[n++] = (stbi_uc) c; + else { + stbi_uc fill = 0; + if (c == 16) { + c = stbi__zreceive(a,2)+3; + if (n == 0) return stbi__err("bad codelengths", "Corrupt PNG"); + fill = lencodes[n-1]; + } else if (c == 17) { + c = stbi__zreceive(a,3)+3; + } else if (c == 18) { + c = stbi__zreceive(a,7)+11; + } else { + return stbi__err("bad codelengths", "Corrupt PNG"); + } + if (ntot - n < c) return stbi__err("bad codelengths", "Corrupt PNG"); + memset(lencodes+n, fill, c); + n += c; + } + } + if (n != ntot) return stbi__err("bad codelengths","Corrupt PNG"); + if (!stbi__zbuild_huffman(&a->z_length, lencodes, hlit)) return 0; + if (!stbi__zbuild_huffman(&a->z_distance, lencodes+hlit, hdist)) return 0; + return 1; +} + +static int stbi__parse_uncompressed_block(stbi__zbuf *a) +{ + stbi_uc header[4]; + int len,nlen,k; + if (a->num_bits & 7) + stbi__zreceive(a, a->num_bits & 7); // discard + // drain the bit-packed data into header + k = 0; + while (a->num_bits > 0) { + header[k++] = (stbi_uc) (a->code_buffer & 255); // suppress MSVC run-time check + a->code_buffer >>= 8; + a->num_bits -= 8; + } + if (a->num_bits < 0) return stbi__err("zlib corrupt","Corrupt PNG"); + // now fill header the normal way + while (k < 4) + header[k++] = stbi__zget8(a); + len = header[1] * 256 + header[0]; + nlen = header[3] * 256 + header[2]; + if (nlen != (len ^ 0xffff)) return stbi__err("zlib corrupt","Corrupt PNG"); + if (a->zbuffer + len > a->zbuffer_end) return stbi__err("read past buffer","Corrupt PNG"); + if (a->zout + len > a->zout_end) + if (!stbi__zexpand(a, a->zout, len)) return 0; + memcpy(a->zout, a->zbuffer, len); + a->zbuffer += len; + a->zout += len; + return 1; +} + +static int stbi__parse_zlib_header(stbi__zbuf *a) +{ + int cmf = stbi__zget8(a); + int cm = cmf & 15; + /* int cinfo = cmf >> 4; */ + int flg = stbi__zget8(a); + if (stbi__zeof(a)) return stbi__err("bad zlib header","Corrupt PNG"); // zlib spec + if ((cmf*256+flg) % 31 != 0) return stbi__err("bad zlib header","Corrupt PNG"); // zlib spec + if (flg & 32) return stbi__err("no preset dict","Corrupt PNG"); // preset dictionary not allowed in png + if (cm != 8) return stbi__err("bad compression","Corrupt PNG"); // DEFLATE required for png + // window = 1 << (8 + cinfo)... but who cares, we fully buffer output + return 1; +} + +static const stbi_uc stbi__zdefault_length[STBI__ZNSYMS] = +{ + 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, + 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, + 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, + 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, + 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, + 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, + 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, + 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, 7,7,7,7,7,7,7,7,8,8,8,8,8,8,8,8 +}; +static const stbi_uc stbi__zdefault_distance[32] = +{ + 5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5 +}; +/* +Init algorithm: +{ + int i; // use <= to match clearly with spec + for (i=0; i <= 143; ++i) stbi__zdefault_length[i] = 8; + for ( ; i <= 255; ++i) stbi__zdefault_length[i] = 9; + for ( ; i <= 279; ++i) stbi__zdefault_length[i] = 7; + for ( ; i <= 287; ++i) stbi__zdefault_length[i] = 8; + + for (i=0; i <= 31; ++i) stbi__zdefault_distance[i] = 5; +} +*/ + +static int stbi__parse_zlib(stbi__zbuf *a, int parse_header) +{ + int final, type; + if (parse_header) + if (!stbi__parse_zlib_header(a)) return 0; + a->num_bits = 0; + a->code_buffer = 0; + a->hit_zeof_once = 0; + do { + final = stbi__zreceive(a,1); + type = stbi__zreceive(a,2); + if (type == 0) { + if (!stbi__parse_uncompressed_block(a)) return 0; + } else if (type == 3) { + return 0; + } else { + if (type == 1) { + // use fixed code lengths + if (!stbi__zbuild_huffman(&a->z_length , stbi__zdefault_length , STBI__ZNSYMS)) return 0; + if (!stbi__zbuild_huffman(&a->z_distance, stbi__zdefault_distance, 32)) return 0; + } else { + if (!stbi__compute_huffman_codes(a)) return 0; + } + if (!stbi__parse_huffman_block(a)) return 0; + } + } while (!final); + return 1; +} + +static int stbi__do_zlib(stbi__zbuf *a, char *obuf, int olen, int exp, int parse_header) +{ + a->zout_start = obuf; + a->zout = obuf; + a->zout_end = obuf + olen; + a->z_expandable = exp; + + return stbi__parse_zlib(a, parse_header); +} + +STBIDEF char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen) +{ + stbi__zbuf a; + char *p = (char *) stbi__malloc(initial_size); + if (p == NULL) return NULL; + a.zbuffer = (stbi_uc *) buffer; + a.zbuffer_end = (stbi_uc *) buffer + len; + if (stbi__do_zlib(&a, p, initial_size, 1, 1)) { + if (outlen) *outlen = (int) (a.zout - a.zout_start); + return a.zout_start; + } else { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF char *stbi_zlib_decode_malloc(char const *buffer, int len, int *outlen) +{ + return stbi_zlib_decode_malloc_guesssize(buffer, len, 16384, outlen); +} + +STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, int len, int initial_size, int *outlen, int parse_header) +{ + stbi__zbuf a; + char *p = (char *) stbi__malloc(initial_size); + if (p == NULL) return NULL; + a.zbuffer = (stbi_uc *) buffer; + a.zbuffer_end = (stbi_uc *) buffer + len; + if (stbi__do_zlib(&a, p, initial_size, 1, parse_header)) { + if (outlen) *outlen = (int) (a.zout - a.zout_start); + return a.zout_start; + } else { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF int stbi_zlib_decode_buffer(char *obuffer, int olen, char const *ibuffer, int ilen) +{ + stbi__zbuf a; + a.zbuffer = (stbi_uc *) ibuffer; + a.zbuffer_end = (stbi_uc *) ibuffer + ilen; + if (stbi__do_zlib(&a, obuffer, olen, 0, 1)) + return (int) (a.zout - a.zout_start); + else + return -1; +} + +STBIDEF char *stbi_zlib_decode_noheader_malloc(char const *buffer, int len, int *outlen) +{ + stbi__zbuf a; + char *p = (char *) stbi__malloc(16384); + if (p == NULL) return NULL; + a.zbuffer = (stbi_uc *) buffer; + a.zbuffer_end = (stbi_uc *) buffer+len; + if (stbi__do_zlib(&a, p, 16384, 1, 0)) { + if (outlen) *outlen = (int) (a.zout - a.zout_start); + return a.zout_start; + } else { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen) +{ + stbi__zbuf a; + a.zbuffer = (stbi_uc *) ibuffer; + a.zbuffer_end = (stbi_uc *) ibuffer + ilen; + if (stbi__do_zlib(&a, obuffer, olen, 0, 0)) + return (int) (a.zout - a.zout_start); + else + return -1; +} +#endif + +// public domain "baseline" PNG decoder v0.10 Sean Barrett 2006-11-18 +// simple implementation +// - only 8-bit samples +// - no CRC checking +// - allocates lots of intermediate memory +// - avoids problem of streaming data between subsystems +// - avoids explicit window management +// performance +// - uses stb_zlib, a PD zlib implementation with fast huffman decoding + +#ifndef STBI_NO_PNG +typedef struct +{ + stbi__uint32 length; + stbi__uint32 type; +} stbi__pngchunk; + +static stbi__pngchunk stbi__get_chunk_header(stbi__context *s) +{ + stbi__pngchunk c; + c.length = stbi__get32be(s); + c.type = stbi__get32be(s); + return c; +} + +static int stbi__check_png_header(stbi__context *s) +{ + static const stbi_uc png_sig[8] = { 137,80,78,71,13,10,26,10 }; + int i; + for (i=0; i < 8; ++i) + if (stbi__get8(s) != png_sig[i]) return stbi__err("bad png sig","Not a PNG"); + return 1; +} + +typedef struct +{ + stbi__context *s; + stbi_uc *idata, *expanded, *out; + int depth; +} stbi__png; + + +enum { + STBI__F_none=0, + STBI__F_sub=1, + STBI__F_up=2, + STBI__F_avg=3, + STBI__F_paeth=4, + // synthetic filter used for first scanline to avoid needing a dummy row of 0s + STBI__F_avg_first +}; + +static stbi_uc first_row_filter[5] = +{ + STBI__F_none, + STBI__F_sub, + STBI__F_none, + STBI__F_avg_first, + STBI__F_sub // Paeth with b=c=0 turns out to be equivalent to sub +}; + +static int stbi__paeth(int a, int b, int c) +{ + // This formulation looks very different from the reference in the PNG spec, but is + // actually equivalent and has favorable data dependencies and admits straightforward + // generation of branch-free code, which helps performance significantly. + int thresh = c*3 - (a + b); + int lo = a < b ? a : b; + int hi = a < b ? b : a; + int t0 = (hi <= thresh) ? lo : c; + int t1 = (thresh <= lo) ? hi : t0; + return t1; +} + +static const stbi_uc stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 }; + +// adds an extra all-255 alpha channel +// dest == src is legal +// img_n must be 1 or 3 +static void stbi__create_png_alpha_expand8(stbi_uc *dest, stbi_uc *src, stbi__uint32 x, int img_n) +{ + int i; + // must process data backwards since we allow dest==src + if (img_n == 1) { + for (i=x-1; i >= 0; --i) { + dest[i*2+1] = 255; + dest[i*2+0] = src[i]; + } + } else { + STBI_ASSERT(img_n == 3); + for (i=x-1; i >= 0; --i) { + dest[i*4+3] = 255; + dest[i*4+2] = src[i*3+2]; + dest[i*4+1] = src[i*3+1]; + dest[i*4+0] = src[i*3+0]; + } + } +} + +// create the png data from post-deflated data +static int stbi__create_png_image_raw(stbi__png *a, stbi_uc *raw, stbi__uint32 raw_len, int out_n, stbi__uint32 x, stbi__uint32 y, int depth, int color) +{ + int bytes = (depth == 16 ? 2 : 1); + stbi__context *s = a->s; + stbi__uint32 i,j,stride = x*out_n*bytes; + stbi__uint32 img_len, img_width_bytes; + stbi_uc *filter_buf; + int all_ok = 1; + int k; + int img_n = s->img_n; // copy it into a local for later + + int output_bytes = out_n*bytes; + int filter_bytes = img_n*bytes; + int width = x; + + STBI_ASSERT(out_n == s->img_n || out_n == s->img_n+1); + a->out = (stbi_uc *) stbi__malloc_mad3(x, y, output_bytes, 0); // extra bytes to write off the end into + if (!a->out) return stbi__err("outofmem", "Out of memory"); + + // note: error exits here don't need to clean up a->out individually, + // stbi__do_png always does on error. + if (!stbi__mad3sizes_valid(img_n, x, depth, 7)) return stbi__err("too large", "Corrupt PNG"); + img_width_bytes = (((img_n * x * depth) + 7) >> 3); + if (!stbi__mad2sizes_valid(img_width_bytes, y, img_width_bytes)) return stbi__err("too large", "Corrupt PNG"); + img_len = (img_width_bytes + 1) * y; + + // we used to check for exact match between raw_len and img_len on non-interlaced PNGs, + // but issue #276 reported a PNG in the wild that had extra data at the end (all zeros), + // so just check for raw_len < img_len always. + if (raw_len < img_len) return stbi__err("not enough pixels","Corrupt PNG"); + + // Allocate two scan lines worth of filter workspace buffer. + filter_buf = (stbi_uc *) stbi__malloc_mad2(img_width_bytes, 2, 0); + if (!filter_buf) return stbi__err("outofmem", "Out of memory"); + + // Filtering for low-bit-depth images + if (depth < 8) { + filter_bytes = 1; + width = img_width_bytes; + } + + for (j=0; j < y; ++j) { + // cur/prior filter buffers alternate + stbi_uc *cur = filter_buf + (j & 1)*img_width_bytes; + stbi_uc *prior = filter_buf + (~j & 1)*img_width_bytes; + stbi_uc *dest = a->out + stride*j; + int nk = width * filter_bytes; + int filter = *raw++; + + // check filter type + if (filter > 4) { + all_ok = stbi__err("invalid filter","Corrupt PNG"); + break; + } + + // if first row, use special filter that doesn't sample previous row + if (j == 0) filter = first_row_filter[filter]; + + // perform actual filtering + switch (filter) { + case STBI__F_none: + memcpy(cur, raw, nk); + break; + case STBI__F_sub: + memcpy(cur, raw, filter_bytes); + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + cur[k-filter_bytes]); + break; + case STBI__F_up: + for (k = 0; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + prior[k]); + break; + case STBI__F_avg: + for (k = 0; k < filter_bytes; ++k) + cur[k] = STBI__BYTECAST(raw[k] + (prior[k]>>1)); + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k-filter_bytes])>>1)); + break; + case STBI__F_paeth: + for (k = 0; k < filter_bytes; ++k) + cur[k] = STBI__BYTECAST(raw[k] + prior[k]); // prior[k] == stbi__paeth(0,prior[k],0) + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-filter_bytes], prior[k], prior[k-filter_bytes])); + break; + case STBI__F_avg_first: + memcpy(cur, raw, filter_bytes); + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + (cur[k-filter_bytes] >> 1)); + break; + } + + raw += nk; + + // expand decoded bits in cur to dest, also adding an extra alpha channel if desired + if (depth < 8) { + stbi_uc scale = (color == 0) ? stbi__depth_scale_table[depth] : 1; // scale grayscale values to 0..255 range + stbi_uc *in = cur; + stbi_uc *out = dest; + stbi_uc inb = 0; + stbi__uint32 nsmp = x*img_n; + + // expand bits to bytes first + if (depth == 4) { + for (i=0; i < nsmp; ++i) { + if ((i & 1) == 0) inb = *in++; + *out++ = scale * (inb >> 4); + inb <<= 4; + } + } else if (depth == 2) { + for (i=0; i < nsmp; ++i) { + if ((i & 3) == 0) inb = *in++; + *out++ = scale * (inb >> 6); + inb <<= 2; + } + } else { + STBI_ASSERT(depth == 1); + for (i=0; i < nsmp; ++i) { + if ((i & 7) == 0) inb = *in++; + *out++ = scale * (inb >> 7); + inb <<= 1; + } + } + + // insert alpha=255 values if desired + if (img_n != out_n) + stbi__create_png_alpha_expand8(dest, dest, x, img_n); + } else if (depth == 8) { + if (img_n == out_n) + memcpy(dest, cur, x*img_n); + else + stbi__create_png_alpha_expand8(dest, cur, x, img_n); + } else if (depth == 16) { + // convert the image data from big-endian to platform-native + stbi__uint16 *dest16 = (stbi__uint16*)dest; + stbi__uint32 nsmp = x*img_n; + + if (img_n == out_n) { + for (i = 0; i < nsmp; ++i, ++dest16, cur += 2) + *dest16 = (cur[0] << 8) | cur[1]; + } else { + STBI_ASSERT(img_n+1 == out_n); + if (img_n == 1) { + for (i = 0; i < x; ++i, dest16 += 2, cur += 2) { + dest16[0] = (cur[0] << 8) | cur[1]; + dest16[1] = 0xffff; + } + } else { + STBI_ASSERT(img_n == 3); + for (i = 0; i < x; ++i, dest16 += 4, cur += 6) { + dest16[0] = (cur[0] << 8) | cur[1]; + dest16[1] = (cur[2] << 8) | cur[3]; + dest16[2] = (cur[4] << 8) | cur[5]; + dest16[3] = 0xffff; + } + } + } + } + } + + STBI_FREE(filter_buf); + if (!all_ok) return 0; + + return 1; +} + +static int stbi__create_png_image(stbi__png *a, stbi_uc *image_data, stbi__uint32 image_data_len, int out_n, int depth, int color, int interlaced) +{ + int bytes = (depth == 16 ? 2 : 1); + int out_bytes = out_n * bytes; + stbi_uc *final; + int p; + if (!interlaced) + return stbi__create_png_image_raw(a, image_data, image_data_len, out_n, a->s->img_x, a->s->img_y, depth, color); + + // de-interlacing + final = (stbi_uc *) stbi__malloc_mad3(a->s->img_x, a->s->img_y, out_bytes, 0); + if (!final) return stbi__err("outofmem", "Out of memory"); + for (p=0; p < 7; ++p) { + int xorig[] = { 0,4,0,2,0,1,0 }; + int yorig[] = { 0,0,4,0,2,0,1 }; + int xspc[] = { 8,8,4,4,2,2,1 }; + int yspc[] = { 8,8,8,4,4,2,2 }; + int i,j,x,y; + // pass1_x[4] = 0, pass1_x[5] = 1, pass1_x[12] = 1 + x = (a->s->img_x - xorig[p] + xspc[p]-1) / xspc[p]; + y = (a->s->img_y - yorig[p] + yspc[p]-1) / yspc[p]; + if (x && y) { + stbi__uint32 img_len = ((((a->s->img_n * x * depth) + 7) >> 3) + 1) * y; + if (!stbi__create_png_image_raw(a, image_data, image_data_len, out_n, x, y, depth, color)) { + STBI_FREE(final); + return 0; + } + for (j=0; j < y; ++j) { + for (i=0; i < x; ++i) { + int out_y = j*yspc[p]+yorig[p]; + int out_x = i*xspc[p]+xorig[p]; + memcpy(final + out_y*a->s->img_x*out_bytes + out_x*out_bytes, + a->out + (j*x+i)*out_bytes, out_bytes); + } + } + STBI_FREE(a->out); + image_data += img_len; + image_data_len -= img_len; + } + } + a->out = final; + + return 1; +} + +static int stbi__compute_transparency(stbi__png *z, stbi_uc tc[3], int out_n) +{ + stbi__context *s = z->s; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi_uc *p = z->out; + + // compute color-based transparency, assuming we've + // already got 255 as the alpha value in the output + STBI_ASSERT(out_n == 2 || out_n == 4); + + if (out_n == 2) { + for (i=0; i < pixel_count; ++i) { + p[1] = (p[0] == tc[0] ? 0 : 255); + p += 2; + } + } else { + for (i=0; i < pixel_count; ++i) { + if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2]) + p[3] = 0; + p += 4; + } + } + return 1; +} + +static int stbi__compute_transparency16(stbi__png *z, stbi__uint16 tc[3], int out_n) +{ + stbi__context *s = z->s; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi__uint16 *p = (stbi__uint16*) z->out; + + // compute color-based transparency, assuming we've + // already got 65535 as the alpha value in the output + STBI_ASSERT(out_n == 2 || out_n == 4); + + if (out_n == 2) { + for (i = 0; i < pixel_count; ++i) { + p[1] = (p[0] == tc[0] ? 0 : 65535); + p += 2; + } + } else { + for (i = 0; i < pixel_count; ++i) { + if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2]) + p[3] = 0; + p += 4; + } + } + return 1; +} + +static int stbi__expand_png_palette(stbi__png *a, stbi_uc *palette, int len, int pal_img_n) +{ + stbi__uint32 i, pixel_count = a->s->img_x * a->s->img_y; + stbi_uc *p, *temp_out, *orig = a->out; + + p = (stbi_uc *) stbi__malloc_mad2(pixel_count, pal_img_n, 0); + if (p == NULL) return stbi__err("outofmem", "Out of memory"); + + // between here and free(out) below, exitting would leak + temp_out = p; + + if (pal_img_n == 3) { + for (i=0; i < pixel_count; ++i) { + int n = orig[i]*4; + p[0] = palette[n ]; + p[1] = palette[n+1]; + p[2] = palette[n+2]; + p += 3; + } + } else { + for (i=0; i < pixel_count; ++i) { + int n = orig[i]*4; + p[0] = palette[n ]; + p[1] = palette[n+1]; + p[2] = palette[n+2]; + p[3] = palette[n+3]; + p += 4; + } + } + STBI_FREE(a->out); + a->out = temp_out; + + STBI_NOTUSED(len); + + return 1; +} + +static int stbi__unpremultiply_on_load_global = 0; +static int stbi__de_iphone_flag_global = 0; + +STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply) +{ + stbi__unpremultiply_on_load_global = flag_true_if_should_unpremultiply; +} + +STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert) +{ + stbi__de_iphone_flag_global = flag_true_if_should_convert; +} + +#ifndef STBI_THREAD_LOCAL +#define stbi__unpremultiply_on_load stbi__unpremultiply_on_load_global +#define stbi__de_iphone_flag stbi__de_iphone_flag_global +#else +static STBI_THREAD_LOCAL int stbi__unpremultiply_on_load_local, stbi__unpremultiply_on_load_set; +static STBI_THREAD_LOCAL int stbi__de_iphone_flag_local, stbi__de_iphone_flag_set; + +STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply) +{ + stbi__unpremultiply_on_load_local = flag_true_if_should_unpremultiply; + stbi__unpremultiply_on_load_set = 1; +} + +STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert) +{ + stbi__de_iphone_flag_local = flag_true_if_should_convert; + stbi__de_iphone_flag_set = 1; +} + +#define stbi__unpremultiply_on_load (stbi__unpremultiply_on_load_set \ + ? stbi__unpremultiply_on_load_local \ + : stbi__unpremultiply_on_load_global) +#define stbi__de_iphone_flag (stbi__de_iphone_flag_set \ + ? stbi__de_iphone_flag_local \ + : stbi__de_iphone_flag_global) +#endif // STBI_THREAD_LOCAL + +static void stbi__de_iphone(stbi__png *z) +{ + stbi__context *s = z->s; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi_uc *p = z->out; + + if (s->img_out_n == 3) { // convert bgr to rgb + for (i=0; i < pixel_count; ++i) { + stbi_uc t = p[0]; + p[0] = p[2]; + p[2] = t; + p += 3; + } + } else { + STBI_ASSERT(s->img_out_n == 4); + if (stbi__unpremultiply_on_load) { + // convert bgr to rgb and unpremultiply + for (i=0; i < pixel_count; ++i) { + stbi_uc a = p[3]; + stbi_uc t = p[0]; + if (a) { + stbi_uc half = a / 2; + p[0] = (p[2] * 255 + half) / a; + p[1] = (p[1] * 255 + half) / a; + p[2] = ( t * 255 + half) / a; + } else { + p[0] = p[2]; + p[2] = t; + } + p += 4; + } + } else { + // convert bgr to rgb + for (i=0; i < pixel_count; ++i) { + stbi_uc t = p[0]; + p[0] = p[2]; + p[2] = t; + p += 4; + } + } + } +} + +#define STBI__PNG_TYPE(a,b,c,d) (((unsigned) (a) << 24) + ((unsigned) (b) << 16) + ((unsigned) (c) << 8) + (unsigned) (d)) + +static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) +{ + stbi_uc palette[1024], pal_img_n=0; + stbi_uc has_trans=0, tc[3]={0}; + stbi__uint16 tc16[3]; + stbi__uint32 ioff=0, idata_limit=0, i, pal_len=0; + int first=1,k,interlace=0, color=0, is_iphone=0; + stbi__context *s = z->s; + + z->expanded = NULL; + z->idata = NULL; + z->out = NULL; + + if (!stbi__check_png_header(s)) return 0; + + if (scan == STBI__SCAN_type) return 1; + + for (;;) { + stbi__pngchunk c = stbi__get_chunk_header(s); + switch (c.type) { + case STBI__PNG_TYPE('C','g','B','I'): + is_iphone = 1; + stbi__skip(s, c.length); + break; + case STBI__PNG_TYPE('I','H','D','R'): { + int comp,filter; + if (!first) return stbi__err("multiple IHDR","Corrupt PNG"); + first = 0; + if (c.length != 13) return stbi__err("bad IHDR len","Corrupt PNG"); + s->img_x = stbi__get32be(s); + s->img_y = stbi__get32be(s); + if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)"); + z->depth = stbi__get8(s); if (z->depth != 1 && z->depth != 2 && z->depth != 4 && z->depth != 8 && z->depth != 16) return stbi__err("1/2/4/8/16-bit only","PNG not supported: 1/2/4/8/16-bit only"); + color = stbi__get8(s); if (color > 6) return stbi__err("bad ctype","Corrupt PNG"); + if (color == 3 && z->depth == 16) return stbi__err("bad ctype","Corrupt PNG"); + if (color == 3) pal_img_n = 3; else if (color & 1) return stbi__err("bad ctype","Corrupt PNG"); + comp = stbi__get8(s); if (comp) return stbi__err("bad comp method","Corrupt PNG"); + filter= stbi__get8(s); if (filter) return stbi__err("bad filter method","Corrupt PNG"); + interlace = stbi__get8(s); if (interlace>1) return stbi__err("bad interlace method","Corrupt PNG"); + if (!s->img_x || !s->img_y) return stbi__err("0-pixel image","Corrupt PNG"); + if (!pal_img_n) { + s->img_n = (color & 2 ? 3 : 1) + (color & 4 ? 1 : 0); + if ((1 << 30) / s->img_x / s->img_n < s->img_y) return stbi__err("too large", "Image too large to decode"); + } else { + // if paletted, then pal_n is our final components, and + // img_n is # components to decompress/filter. + s->img_n = 1; + if ((1 << 30) / s->img_x / 4 < s->img_y) return stbi__err("too large","Corrupt PNG"); + } + // even with SCAN_header, have to scan to see if we have a tRNS + break; + } + + case STBI__PNG_TYPE('P','L','T','E'): { + if (first) return stbi__err("first not IHDR", "Corrupt PNG"); + if (c.length > 256*3) return stbi__err("invalid PLTE","Corrupt PNG"); + pal_len = c.length / 3; + if (pal_len * 3 != c.length) return stbi__err("invalid PLTE","Corrupt PNG"); + for (i=0; i < pal_len; ++i) { + palette[i*4+0] = stbi__get8(s); + palette[i*4+1] = stbi__get8(s); + palette[i*4+2] = stbi__get8(s); + palette[i*4+3] = 255; + } + break; + } + + case STBI__PNG_TYPE('t','R','N','S'): { + if (first) return stbi__err("first not IHDR", "Corrupt PNG"); + if (z->idata) return stbi__err("tRNS after IDAT","Corrupt PNG"); + if (pal_img_n) { + if (scan == STBI__SCAN_header) { s->img_n = 4; return 1; } + if (pal_len == 0) return stbi__err("tRNS before PLTE","Corrupt PNG"); + if (c.length > pal_len) return stbi__err("bad tRNS len","Corrupt PNG"); + pal_img_n = 4; + for (i=0; i < c.length; ++i) + palette[i*4+3] = stbi__get8(s); + } else { + if (!(s->img_n & 1)) return stbi__err("tRNS with alpha","Corrupt PNG"); + if (c.length != (stbi__uint32) s->img_n*2) return stbi__err("bad tRNS len","Corrupt PNG"); + has_trans = 1; + // non-paletted with tRNS = constant alpha. if header-scanning, we can stop now. + if (scan == STBI__SCAN_header) { ++s->img_n; return 1; } + if (z->depth == 16) { + for (k = 0; k < s->img_n && k < 3; ++k) // extra loop test to suppress false GCC warning + tc16[k] = (stbi__uint16)stbi__get16be(s); // copy the values as-is + } else { + for (k = 0; k < s->img_n && k < 3; ++k) + tc[k] = (stbi_uc)(stbi__get16be(s) & 255) * stbi__depth_scale_table[z->depth]; // non 8-bit images will be larger + } + } + break; + } + + case STBI__PNG_TYPE('I','D','A','T'): { + if (first) return stbi__err("first not IHDR", "Corrupt PNG"); + if (pal_img_n && !pal_len) return stbi__err("no PLTE","Corrupt PNG"); + if (scan == STBI__SCAN_header) { + // header scan definitely stops at first IDAT + if (pal_img_n) + s->img_n = pal_img_n; + return 1; + } + if (c.length > (1u << 30)) return stbi__err("IDAT size limit", "IDAT section larger than 2^30 bytes"); + if ((int)(ioff + c.length) < (int)ioff) return 0; + if (ioff + c.length > idata_limit) { + stbi__uint32 idata_limit_old = idata_limit; + stbi_uc *p; + if (idata_limit == 0) idata_limit = c.length > 4096 ? c.length : 4096; + while (ioff + c.length > idata_limit) + idata_limit *= 2; + STBI_NOTUSED(idata_limit_old); + p = (stbi_uc *) STBI_REALLOC_SIZED(z->idata, idata_limit_old, idata_limit); if (p == NULL) return stbi__err("outofmem", "Out of memory"); + z->idata = p; + } + if (!stbi__getn(s, z->idata+ioff,c.length)) return stbi__err("outofdata","Corrupt PNG"); + ioff += c.length; + break; + } + + case STBI__PNG_TYPE('I','E','N','D'): { + stbi__uint32 raw_len, bpl; + if (first) return stbi__err("first not IHDR", "Corrupt PNG"); + if (scan != STBI__SCAN_load) return 1; + if (z->idata == NULL) return stbi__err("no IDAT","Corrupt PNG"); + // initial guess for decoded data size to avoid unnecessary reallocs + bpl = (s->img_x * z->depth + 7) / 8; // bytes per line, per component + raw_len = bpl * s->img_y * s->img_n /* pixels */ + s->img_y /* filter mode per row */; + z->expanded = (stbi_uc *) stbi_zlib_decode_malloc_guesssize_headerflag((char *) z->idata, ioff, raw_len, (int *) &raw_len, !is_iphone); + if (z->expanded == NULL) return 0; // zlib should set error + STBI_FREE(z->idata); z->idata = NULL; + if ((req_comp == s->img_n+1 && req_comp != 3 && !pal_img_n) || has_trans) + s->img_out_n = s->img_n+1; + else + s->img_out_n = s->img_n; + if (!stbi__create_png_image(z, z->expanded, raw_len, s->img_out_n, z->depth, color, interlace)) return 0; + if (has_trans) { + if (z->depth == 16) { + if (!stbi__compute_transparency16(z, tc16, s->img_out_n)) return 0; + } else { + if (!stbi__compute_transparency(z, tc, s->img_out_n)) return 0; + } + } + if (is_iphone && stbi__de_iphone_flag && s->img_out_n > 2) + stbi__de_iphone(z); + if (pal_img_n) { + // pal_img_n == 3 or 4 + s->img_n = pal_img_n; // record the actual colors we had + s->img_out_n = pal_img_n; + if (req_comp >= 3) s->img_out_n = req_comp; + if (!stbi__expand_png_palette(z, palette, pal_len, s->img_out_n)) + return 0; + } else if (has_trans) { + // non-paletted image with tRNS -> source image has (constant) alpha + ++s->img_n; + } + STBI_FREE(z->expanded); z->expanded = NULL; + // end of PNG chunk, read and skip CRC + stbi__get32be(s); + return 1; + } + + default: + // if critical, fail + if (first) return stbi__err("first not IHDR", "Corrupt PNG"); + if ((c.type & (1 << 29)) == 0) { + #ifndef STBI_NO_FAILURE_STRINGS + // not threadsafe + static char invalid_chunk[] = "XXXX PNG chunk not known"; + invalid_chunk[0] = STBI__BYTECAST(c.type >> 24); + invalid_chunk[1] = STBI__BYTECAST(c.type >> 16); + invalid_chunk[2] = STBI__BYTECAST(c.type >> 8); + invalid_chunk[3] = STBI__BYTECAST(c.type >> 0); + #endif + return stbi__err(invalid_chunk, "PNG not supported: unknown PNG chunk type"); + } + stbi__skip(s, c.length); + break; + } + // end of PNG chunk, read and skip CRC + stbi__get32be(s); + } +} + +static void *stbi__do_png(stbi__png *p, int *x, int *y, int *n, int req_comp, stbi__result_info *ri) +{ + void *result=NULL; + if (req_comp < 0 || req_comp > 4) return stbi__errpuc("bad req_comp", "Internal error"); + if (stbi__parse_png_file(p, STBI__SCAN_load, req_comp)) { + if (p->depth <= 8) + ri->bits_per_channel = 8; + else if (p->depth == 16) + ri->bits_per_channel = 16; + else + return stbi__errpuc("bad bits_per_channel", "PNG not supported: unsupported color depth"); + result = p->out; + p->out = NULL; + if (req_comp && req_comp != p->s->img_out_n) { + if (ri->bits_per_channel == 8) + result = stbi__convert_format((unsigned char *) result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y); + else + result = stbi__convert_format16((stbi__uint16 *) result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y); + p->s->img_out_n = req_comp; + if (result == NULL) return result; + } + *x = p->s->img_x; + *y = p->s->img_y; + if (n) *n = p->s->img_n; + } + STBI_FREE(p->out); p->out = NULL; + STBI_FREE(p->expanded); p->expanded = NULL; + STBI_FREE(p->idata); p->idata = NULL; + + return result; +} + +static void *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri) +{ + stbi__png p; + p.s = s; + return stbi__do_png(&p, x,y,comp,req_comp, ri); +} + +static int stbi__png_test(stbi__context *s) +{ + int r; + r = stbi__check_png_header(s); + stbi__rewind(s); + return r; +} + +static int stbi__png_info_raw(stbi__png *p, int *x, int *y, int *comp) +{ + if (!stbi__parse_png_file(p, STBI__SCAN_header, 0)) { + stbi__rewind( p->s ); + return 0; + } + if (x) *x = p->s->img_x; + if (y) *y = p->s->img_y; + if (comp) *comp = p->s->img_n; + return 1; +} + +static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp) +{ + stbi__png p; + p.s = s; + return stbi__png_info_raw(&p, x, y, comp); +} + +static int stbi__png_is16(stbi__context *s) +{ + stbi__png p; + p.s = s; + if (!stbi__png_info_raw(&p, NULL, NULL, NULL)) + return 0; + if (p.depth != 16) { + stbi__rewind(p.s); + return 0; + } + return 1; +} +#endif + +// Microsoft/Windows BMP image + +#ifndef STBI_NO_BMP +static int stbi__bmp_test_raw(stbi__context *s) +{ + int r; + int sz; + if (stbi__get8(s) != 'B') return 0; + if (stbi__get8(s) != 'M') return 0; + stbi__get32le(s); // discard filesize + stbi__get16le(s); // discard reserved + stbi__get16le(s); // discard reserved + stbi__get32le(s); // discard data offset + sz = stbi__get32le(s); + r = (sz == 12 || sz == 40 || sz == 56 || sz == 108 || sz == 124); + return r; +} + +static int stbi__bmp_test(stbi__context *s) +{ + int r = stbi__bmp_test_raw(s); + stbi__rewind(s); + return r; +} + + +// returns 0..31 for the highest set bit +static int stbi__high_bit(unsigned int z) +{ + int n=0; + if (z == 0) return -1; + if (z >= 0x10000) { n += 16; z >>= 16; } + if (z >= 0x00100) { n += 8; z >>= 8; } + if (z >= 0x00010) { n += 4; z >>= 4; } + if (z >= 0x00004) { n += 2; z >>= 2; } + if (z >= 0x00002) { n += 1;/* >>= 1;*/ } + return n; +} + +static int stbi__bitcount(unsigned int a) +{ + a = (a & 0x55555555) + ((a >> 1) & 0x55555555); // max 2 + a = (a & 0x33333333) + ((a >> 2) & 0x33333333); // max 4 + a = (a + (a >> 4)) & 0x0f0f0f0f; // max 8 per 4, now 8 bits + a = (a + (a >> 8)); // max 16 per 8 bits + a = (a + (a >> 16)); // max 32 per 8 bits + return a & 0xff; +} + +// extract an arbitrarily-aligned N-bit value (N=bits) +// from v, and then make it 8-bits long and fractionally +// extend it to full full range. +static int stbi__shiftsigned(unsigned int v, int shift, int bits) +{ + static unsigned int mul_table[9] = { + 0, + 0xff/*0b11111111*/, 0x55/*0b01010101*/, 0x49/*0b01001001*/, 0x11/*0b00010001*/, + 0x21/*0b00100001*/, 0x41/*0b01000001*/, 0x81/*0b10000001*/, 0x01/*0b00000001*/, + }; + static unsigned int shift_table[9] = { + 0, 0,0,1,0,2,4,6,0, + }; + if (shift < 0) + v <<= -shift; + else + v >>= shift; + STBI_ASSERT(v < 256); + v >>= (8-bits); + STBI_ASSERT(bits >= 0 && bits <= 8); + return (int) ((unsigned) v * mul_table[bits]) >> shift_table[bits]; +} + +typedef struct +{ + int bpp, offset, hsz; + unsigned int mr,mg,mb,ma, all_a; + int extra_read; +} stbi__bmp_data; + +static int stbi__bmp_set_mask_defaults(stbi__bmp_data *info, int compress) +{ + // BI_BITFIELDS specifies masks explicitly, don't override + if (compress == 3) + return 1; + + if (compress == 0) { + if (info->bpp == 16) { + info->mr = 31u << 10; + info->mg = 31u << 5; + info->mb = 31u << 0; + } else if (info->bpp == 32) { + info->mr = 0xffu << 16; + info->mg = 0xffu << 8; + info->mb = 0xffu << 0; + info->ma = 0xffu << 24; + info->all_a = 0; // if all_a is 0 at end, then we loaded alpha channel but it was all 0 + } else { + // otherwise, use defaults, which is all-0 + info->mr = info->mg = info->mb = info->ma = 0; + } + return 1; + } + return 0; // error +} + +static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info) +{ + int hsz; + if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') return stbi__errpuc("not BMP", "Corrupt BMP"); + stbi__get32le(s); // discard filesize + stbi__get16le(s); // discard reserved + stbi__get16le(s); // discard reserved + info->offset = stbi__get32le(s); + info->hsz = hsz = stbi__get32le(s); + info->mr = info->mg = info->mb = info->ma = 0; + info->extra_read = 14; + + if (info->offset < 0) return stbi__errpuc("bad BMP", "bad BMP"); + + if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) return stbi__errpuc("unknown BMP", "BMP type not supported: unknown"); + if (hsz == 12) { + s->img_x = stbi__get16le(s); + s->img_y = stbi__get16le(s); + } else { + s->img_x = stbi__get32le(s); + s->img_y = stbi__get32le(s); + } + if (stbi__get16le(s) != 1) return stbi__errpuc("bad BMP", "bad BMP"); + info->bpp = stbi__get16le(s); + if (hsz != 12) { + int compress = stbi__get32le(s); + if (compress == 1 || compress == 2) return stbi__errpuc("BMP RLE", "BMP type not supported: RLE"); + if (compress >= 4) return stbi__errpuc("BMP JPEG/PNG", "BMP type not supported: unsupported compression"); // this includes PNG/JPEG modes + if (compress == 3 && info->bpp != 16 && info->bpp != 32) return stbi__errpuc("bad BMP", "bad BMP"); // bitfields requires 16 or 32 bits/pixel + stbi__get32le(s); // discard sizeof + stbi__get32le(s); // discard hres + stbi__get32le(s); // discard vres + stbi__get32le(s); // discard colorsused + stbi__get32le(s); // discard max important + if (hsz == 40 || hsz == 56) { + if (hsz == 56) { + stbi__get32le(s); + stbi__get32le(s); + stbi__get32le(s); + stbi__get32le(s); + } + if (info->bpp == 16 || info->bpp == 32) { + if (compress == 0) { + stbi__bmp_set_mask_defaults(info, compress); + } else if (compress == 3) { + info->mr = stbi__get32le(s); + info->mg = stbi__get32le(s); + info->mb = stbi__get32le(s); + info->extra_read += 12; + // not documented, but generated by photoshop and handled by mspaint + if (info->mr == info->mg && info->mg == info->mb) { + // ?!?!? + return stbi__errpuc("bad BMP", "bad BMP"); + } + } else + return stbi__errpuc("bad BMP", "bad BMP"); + } + } else { + // V4/V5 header + int i; + if (hsz != 108 && hsz != 124) + return stbi__errpuc("bad BMP", "bad BMP"); + info->mr = stbi__get32le(s); + info->mg = stbi__get32le(s); + info->mb = stbi__get32le(s); + info->ma = stbi__get32le(s); + if (compress != 3) // override mr/mg/mb unless in BI_BITFIELDS mode, as per docs + stbi__bmp_set_mask_defaults(info, compress); + stbi__get32le(s); // discard color space + for (i=0; i < 12; ++i) + stbi__get32le(s); // discard color space parameters + if (hsz == 124) { + stbi__get32le(s); // discard rendering intent + stbi__get32le(s); // discard offset of profile data + stbi__get32le(s); // discard size of profile data + stbi__get32le(s); // discard reserved + } + } + } + return (void *) 1; +} + + +static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri) +{ + stbi_uc *out; + unsigned int mr=0,mg=0,mb=0,ma=0, all_a; + stbi_uc pal[256][4]; + int psize=0,i,j,width; + int flip_vertically, pad, target; + stbi__bmp_data info; + STBI_NOTUSED(ri); + + info.all_a = 255; + if (stbi__bmp_parse_header(s, &info) == NULL) + return NULL; // error code already set + + flip_vertically = ((int) s->img_y) > 0; + s->img_y = abs((int) s->img_y); + + if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + + mr = info.mr; + mg = info.mg; + mb = info.mb; + ma = info.ma; + all_a = info.all_a; + + if (info.hsz == 12) { + if (info.bpp < 24) + psize = (info.offset - info.extra_read - 24) / 3; + } else { + if (info.bpp < 16) + psize = (info.offset - info.extra_read - info.hsz) >> 2; + } + if (psize == 0) { + // accept some number of extra bytes after the header, but if the offset points either to before + // the header ends or implies a large amount of extra data, reject the file as malformed + int bytes_read_so_far = s->callback_already_read + (int)(s->img_buffer - s->img_buffer_original); + int header_limit = 1024; // max we actually read is below 256 bytes currently. + int extra_data_limit = 256*4; // what ordinarily goes here is a palette; 256 entries*4 bytes is its max size. + if (bytes_read_so_far <= 0 || bytes_read_so_far > header_limit) { + return stbi__errpuc("bad header", "Corrupt BMP"); + } + // we established that bytes_read_so_far is positive and sensible. + // the first half of this test rejects offsets that are either too small positives, or + // negative, and guarantees that info.offset >= bytes_read_so_far > 0. this in turn + // ensures the number computed in the second half of the test can't overflow. + if (info.offset < bytes_read_so_far || info.offset - bytes_read_so_far > extra_data_limit) { + return stbi__errpuc("bad offset", "Corrupt BMP"); + } else { + stbi__skip(s, info.offset - bytes_read_so_far); + } + } + + if (info.bpp == 24 && ma == 0xff000000) + s->img_n = 3; + else + s->img_n = ma ? 4 : 3; + if (req_comp && req_comp >= 3) // we can directly decode 3 or 4 + target = req_comp; + else + target = s->img_n; // if they want monochrome, we'll post-convert + + // sanity-check size + if (!stbi__mad3sizes_valid(target, s->img_x, s->img_y, 0)) + return stbi__errpuc("too large", "Corrupt BMP"); + + out = (stbi_uc *) stbi__malloc_mad3(target, s->img_x, s->img_y, 0); + if (!out) return stbi__errpuc("outofmem", "Out of memory"); + if (info.bpp < 16) { + int z=0; + if (psize == 0 || psize > 256) { STBI_FREE(out); return stbi__errpuc("invalid", "Corrupt BMP"); } + for (i=0; i < psize; ++i) { + pal[i][2] = stbi__get8(s); + pal[i][1] = stbi__get8(s); + pal[i][0] = stbi__get8(s); + if (info.hsz != 12) stbi__get8(s); + pal[i][3] = 255; + } + stbi__skip(s, info.offset - info.extra_read - info.hsz - psize * (info.hsz == 12 ? 3 : 4)); + if (info.bpp == 1) width = (s->img_x + 7) >> 3; + else if (info.bpp == 4) width = (s->img_x + 1) >> 1; + else if (info.bpp == 8) width = s->img_x; + else { STBI_FREE(out); return stbi__errpuc("bad bpp", "Corrupt BMP"); } + pad = (-width)&3; + if (info.bpp == 1) { + for (j=0; j < (int) s->img_y; ++j) { + int bit_offset = 7, v = stbi__get8(s); + for (i=0; i < (int) s->img_x; ++i) { + int color = (v>>bit_offset)&0x1; + out[z++] = pal[color][0]; + out[z++] = pal[color][1]; + out[z++] = pal[color][2]; + if (target == 4) out[z++] = 255; + if (i+1 == (int) s->img_x) break; + if((--bit_offset) < 0) { + bit_offset = 7; + v = stbi__get8(s); + } + } + stbi__skip(s, pad); + } + } else { + for (j=0; j < (int) s->img_y; ++j) { + for (i=0; i < (int) s->img_x; i += 2) { + int v=stbi__get8(s),v2=0; + if (info.bpp == 4) { + v2 = v & 15; + v >>= 4; + } + out[z++] = pal[v][0]; + out[z++] = pal[v][1]; + out[z++] = pal[v][2]; + if (target == 4) out[z++] = 255; + if (i+1 == (int) s->img_x) break; + v = (info.bpp == 8) ? stbi__get8(s) : v2; + out[z++] = pal[v][0]; + out[z++] = pal[v][1]; + out[z++] = pal[v][2]; + if (target == 4) out[z++] = 255; + } + stbi__skip(s, pad); + } + } + } else { + int rshift=0,gshift=0,bshift=0,ashift=0,rcount=0,gcount=0,bcount=0,acount=0; + int z = 0; + int easy=0; + stbi__skip(s, info.offset - info.extra_read - info.hsz); + if (info.bpp == 24) width = 3 * s->img_x; + else if (info.bpp == 16) width = 2*s->img_x; + else /* bpp = 32 and pad = 0 */ width=0; + pad = (-width) & 3; + if (info.bpp == 24) { + easy = 1; + } else if (info.bpp == 32) { + if (mb == 0xff && mg == 0xff00 && mr == 0x00ff0000 && ma == 0xff000000) + easy = 2; + } + if (!easy) { + if (!mr || !mg || !mb) { STBI_FREE(out); return stbi__errpuc("bad masks", "Corrupt BMP"); } + // right shift amt to put high bit in position #7 + rshift = stbi__high_bit(mr)-7; rcount = stbi__bitcount(mr); + gshift = stbi__high_bit(mg)-7; gcount = stbi__bitcount(mg); + bshift = stbi__high_bit(mb)-7; bcount = stbi__bitcount(mb); + ashift = stbi__high_bit(ma)-7; acount = stbi__bitcount(ma); + if (rcount > 8 || gcount > 8 || bcount > 8 || acount > 8) { STBI_FREE(out); return stbi__errpuc("bad masks", "Corrupt BMP"); } + } + for (j=0; j < (int) s->img_y; ++j) { + if (easy) { + for (i=0; i < (int) s->img_x; ++i) { + unsigned char a; + out[z+2] = stbi__get8(s); + out[z+1] = stbi__get8(s); + out[z+0] = stbi__get8(s); + z += 3; + a = (easy == 2 ? stbi__get8(s) : 255); + all_a |= a; + if (target == 4) out[z++] = a; + } + } else { + int bpp = info.bpp; + for (i=0; i < (int) s->img_x; ++i) { + stbi__uint32 v = (bpp == 16 ? (stbi__uint32) stbi__get16le(s) : stbi__get32le(s)); + unsigned int a; + out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mr, rshift, rcount)); + out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mg, gshift, gcount)); + out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mb, bshift, bcount)); + a = (ma ? stbi__shiftsigned(v & ma, ashift, acount) : 255); + all_a |= a; + if (target == 4) out[z++] = STBI__BYTECAST(a); + } + } + stbi__skip(s, pad); + } + } + + // if alpha channel is all 0s, replace with all 255s + if (target == 4 && all_a == 0) + for (i=4*s->img_x*s->img_y-1; i >= 0; i -= 4) + out[i] = 255; + + if (flip_vertically) { + stbi_uc t; + for (j=0; j < (int) s->img_y>>1; ++j) { + stbi_uc *p1 = out + j *s->img_x*target; + stbi_uc *p2 = out + (s->img_y-1-j)*s->img_x*target; + for (i=0; i < (int) s->img_x*target; ++i) { + t = p1[i]; p1[i] = p2[i]; p2[i] = t; + } + } + } + + if (req_comp && req_comp != target) { + out = stbi__convert_format(out, target, req_comp, s->img_x, s->img_y); + if (out == NULL) return out; // stbi__convert_format frees input on failure + } + + *x = s->img_x; + *y = s->img_y; + if (comp) *comp = s->img_n; + return out; +} +#endif + +// Targa Truevision - TGA +// by Jonathan Dummer +#ifndef STBI_NO_TGA +// returns STBI_rgb or whatever, 0 on error +static int stbi__tga_get_comp(int bits_per_pixel, int is_grey, int* is_rgb16) +{ + // only RGB or RGBA (incl. 16bit) or grey allowed + if (is_rgb16) *is_rgb16 = 0; + switch(bits_per_pixel) { + case 8: return STBI_grey; + case 16: if(is_grey) return STBI_grey_alpha; + // fallthrough + case 15: if(is_rgb16) *is_rgb16 = 1; + return STBI_rgb; + case 24: // fallthrough + case 32: return bits_per_pixel/8; + default: return 0; + } +} + +static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp) +{ + int tga_w, tga_h, tga_comp, tga_image_type, tga_bits_per_pixel, tga_colormap_bpp; + int sz, tga_colormap_type; + stbi__get8(s); // discard Offset + tga_colormap_type = stbi__get8(s); // colormap type + if( tga_colormap_type > 1 ) { + stbi__rewind(s); + return 0; // only RGB or indexed allowed + } + tga_image_type = stbi__get8(s); // image type + if ( tga_colormap_type == 1 ) { // colormapped (paletted) image + if (tga_image_type != 1 && tga_image_type != 9) { + stbi__rewind(s); + return 0; + } + stbi__skip(s,4); // skip index of first colormap entry and number of entries + sz = stbi__get8(s); // check bits per palette color entry + if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) { + stbi__rewind(s); + return 0; + } + stbi__skip(s,4); // skip image x and y origin + tga_colormap_bpp = sz; + } else { // "normal" image w/o colormap - only RGB or grey allowed, +/- RLE + if ( (tga_image_type != 2) && (tga_image_type != 3) && (tga_image_type != 10) && (tga_image_type != 11) ) { + stbi__rewind(s); + return 0; // only RGB or grey allowed, +/- RLE + } + stbi__skip(s,9); // skip colormap specification and image x/y origin + tga_colormap_bpp = 0; + } + tga_w = stbi__get16le(s); + if( tga_w < 1 ) { + stbi__rewind(s); + return 0; // test width + } + tga_h = stbi__get16le(s); + if( tga_h < 1 ) { + stbi__rewind(s); + return 0; // test height + } + tga_bits_per_pixel = stbi__get8(s); // bits per pixel + stbi__get8(s); // ignore alpha bits + if (tga_colormap_bpp != 0) { + if((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16)) { + // when using a colormap, tga_bits_per_pixel is the size of the indexes + // I don't think anything but 8 or 16bit indexes makes sense + stbi__rewind(s); + return 0; + } + tga_comp = stbi__tga_get_comp(tga_colormap_bpp, 0, NULL); + } else { + tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3) || (tga_image_type == 11), NULL); + } + if(!tga_comp) { + stbi__rewind(s); + return 0; + } + if (x) *x = tga_w; + if (y) *y = tga_h; + if (comp) *comp = tga_comp; + return 1; // seems to have passed everything +} + +static int stbi__tga_test(stbi__context *s) +{ + int res = 0; + int sz, tga_color_type; + stbi__get8(s); // discard Offset + tga_color_type = stbi__get8(s); // color type + if ( tga_color_type > 1 ) goto errorEnd; // only RGB or indexed allowed + sz = stbi__get8(s); // image type + if ( tga_color_type == 1 ) { // colormapped (paletted) image + if (sz != 1 && sz != 9) goto errorEnd; // colortype 1 demands image type 1 or 9 + stbi__skip(s,4); // skip index of first colormap entry and number of entries + sz = stbi__get8(s); // check bits per palette color entry + if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) goto errorEnd; + stbi__skip(s,4); // skip image x and y origin + } else { // "normal" image w/o colormap + if ( (sz != 2) && (sz != 3) && (sz != 10) && (sz != 11) ) goto errorEnd; // only RGB or grey allowed, +/- RLE + stbi__skip(s,9); // skip colormap specification and image x/y origin + } + if ( stbi__get16le(s) < 1 ) goto errorEnd; // test width + if ( stbi__get16le(s) < 1 ) goto errorEnd; // test height + sz = stbi__get8(s); // bits per pixel + if ( (tga_color_type == 1) && (sz != 8) && (sz != 16) ) goto errorEnd; // for colormapped images, bpp is size of an index + if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) goto errorEnd; + + res = 1; // if we got this far, everything's good and we can return 1 instead of 0 + +errorEnd: + stbi__rewind(s); + return res; +} + +// read 16bit value and convert to 24bit RGB +static void stbi__tga_read_rgb16(stbi__context *s, stbi_uc* out) +{ + stbi__uint16 px = (stbi__uint16)stbi__get16le(s); + stbi__uint16 fiveBitMask = 31; + // we have 3 channels with 5bits each + int r = (px >> 10) & fiveBitMask; + int g = (px >> 5) & fiveBitMask; + int b = px & fiveBitMask; + // Note that this saves the data in RGB(A) order, so it doesn't need to be swapped later + out[0] = (stbi_uc)((r * 255)/31); + out[1] = (stbi_uc)((g * 255)/31); + out[2] = (stbi_uc)((b * 255)/31); + + // some people claim that the most significant bit might be used for alpha + // (possibly if an alpha-bit is set in the "image descriptor byte") + // but that only made 16bit test images completely translucent.. + // so let's treat all 15 and 16bit TGAs as RGB with no alpha. +} + +static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri) +{ + // read in the TGA header stuff + int tga_offset = stbi__get8(s); + int tga_indexed = stbi__get8(s); + int tga_image_type = stbi__get8(s); + int tga_is_RLE = 0; + int tga_palette_start = stbi__get16le(s); + int tga_palette_len = stbi__get16le(s); + int tga_palette_bits = stbi__get8(s); + int tga_x_origin = stbi__get16le(s); + int tga_y_origin = stbi__get16le(s); + int tga_width = stbi__get16le(s); + int tga_height = stbi__get16le(s); + int tga_bits_per_pixel = stbi__get8(s); + int tga_comp, tga_rgb16=0; + int tga_inverted = stbi__get8(s); + // int tga_alpha_bits = tga_inverted & 15; // the 4 lowest bits - unused (useless?) + // image data + unsigned char *tga_data; + unsigned char *tga_palette = NULL; + int i, j; + unsigned char raw_data[4] = {0}; + int RLE_count = 0; + int RLE_repeating = 0; + int read_next_pixel = 1; + STBI_NOTUSED(ri); + STBI_NOTUSED(tga_x_origin); // @TODO + STBI_NOTUSED(tga_y_origin); // @TODO + + if (tga_height > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + if (tga_width > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + + // do a tiny bit of precessing + if ( tga_image_type >= 8 ) + { + tga_image_type -= 8; + tga_is_RLE = 1; + } + tga_inverted = 1 - ((tga_inverted >> 5) & 1); + + // If I'm paletted, then I'll use the number of bits from the palette + if ( tga_indexed ) tga_comp = stbi__tga_get_comp(tga_palette_bits, 0, &tga_rgb16); + else tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3), &tga_rgb16); + + if(!tga_comp) // shouldn't really happen, stbi__tga_test() should have ensured basic consistency + return stbi__errpuc("bad format", "Can't find out TGA pixelformat"); + + // tga info + *x = tga_width; + *y = tga_height; + if (comp) *comp = tga_comp; + + if (!stbi__mad3sizes_valid(tga_width, tga_height, tga_comp, 0)) + return stbi__errpuc("too large", "Corrupt TGA"); + + tga_data = (unsigned char*)stbi__malloc_mad3(tga_width, tga_height, tga_comp, 0); + if (!tga_data) return stbi__errpuc("outofmem", "Out of memory"); + + // skip to the data's starting position (offset usually = 0) + stbi__skip(s, tga_offset ); + + if ( !tga_indexed && !tga_is_RLE && !tga_rgb16 ) { + for (i=0; i < tga_height; ++i) { + int row = tga_inverted ? tga_height -i - 1 : i; + stbi_uc *tga_row = tga_data + row*tga_width*tga_comp; + stbi__getn(s, tga_row, tga_width * tga_comp); + } + } else { + // do I need to load a palette? + if ( tga_indexed) + { + if (tga_palette_len == 0) { /* you have to have at least one entry! */ + STBI_FREE(tga_data); + return stbi__errpuc("bad palette", "Corrupt TGA"); + } + + // any data to skip? (offset usually = 0) + stbi__skip(s, tga_palette_start ); + // load the palette + tga_palette = (unsigned char*)stbi__malloc_mad2(tga_palette_len, tga_comp, 0); + if (!tga_palette) { + STBI_FREE(tga_data); + return stbi__errpuc("outofmem", "Out of memory"); + } + if (tga_rgb16) { + stbi_uc *pal_entry = tga_palette; + STBI_ASSERT(tga_comp == STBI_rgb); + for (i=0; i < tga_palette_len; ++i) { + stbi__tga_read_rgb16(s, pal_entry); + pal_entry += tga_comp; + } + } else if (!stbi__getn(s, tga_palette, tga_palette_len * tga_comp)) { + STBI_FREE(tga_data); + STBI_FREE(tga_palette); + return stbi__errpuc("bad palette", "Corrupt TGA"); + } + } + // load the data + for (i=0; i < tga_width * tga_height; ++i) + { + // if I'm in RLE mode, do I need to get a RLE stbi__pngchunk? + if ( tga_is_RLE ) + { + if ( RLE_count == 0 ) + { + // yep, get the next byte as a RLE command + int RLE_cmd = stbi__get8(s); + RLE_count = 1 + (RLE_cmd & 127); + RLE_repeating = RLE_cmd >> 7; + read_next_pixel = 1; + } else if ( !RLE_repeating ) + { + read_next_pixel = 1; + } + } else + { + read_next_pixel = 1; + } + // OK, if I need to read a pixel, do it now + if ( read_next_pixel ) + { + // load however much data we did have + if ( tga_indexed ) + { + // read in index, then perform the lookup + int pal_idx = (tga_bits_per_pixel == 8) ? stbi__get8(s) : stbi__get16le(s); + if ( pal_idx >= tga_palette_len ) { + // invalid index + pal_idx = 0; + } + pal_idx *= tga_comp; + for (j = 0; j < tga_comp; ++j) { + raw_data[j] = tga_palette[pal_idx+j]; + } + } else if(tga_rgb16) { + STBI_ASSERT(tga_comp == STBI_rgb); + stbi__tga_read_rgb16(s, raw_data); + } else { + // read in the data raw + for (j = 0; j < tga_comp; ++j) { + raw_data[j] = stbi__get8(s); + } + } + // clear the reading flag for the next pixel + read_next_pixel = 0; + } // end of reading a pixel + + // copy data + for (j = 0; j < tga_comp; ++j) + tga_data[i*tga_comp+j] = raw_data[j]; + + // in case we're in RLE mode, keep counting down + --RLE_count; + } + // do I need to invert the image? + if ( tga_inverted ) + { + for (j = 0; j*2 < tga_height; ++j) + { + int index1 = j * tga_width * tga_comp; + int index2 = (tga_height - 1 - j) * tga_width * tga_comp; + for (i = tga_width * tga_comp; i > 0; --i) + { + unsigned char temp = tga_data[index1]; + tga_data[index1] = tga_data[index2]; + tga_data[index2] = temp; + ++index1; + ++index2; + } + } + } + // clear my palette, if I had one + if ( tga_palette != NULL ) + { + STBI_FREE( tga_palette ); + } + } + + // swap RGB - if the source data was RGB16, it already is in the right order + if (tga_comp >= 3 && !tga_rgb16) + { + unsigned char* tga_pixel = tga_data; + for (i=0; i < tga_width * tga_height; ++i) + { + unsigned char temp = tga_pixel[0]; + tga_pixel[0] = tga_pixel[2]; + tga_pixel[2] = temp; + tga_pixel += tga_comp; + } + } + + // convert to target component count + if (req_comp && req_comp != tga_comp) + tga_data = stbi__convert_format(tga_data, tga_comp, req_comp, tga_width, tga_height); + + // the things I do to get rid of an error message, and yet keep + // Microsoft's C compilers happy... [8^( + tga_palette_start = tga_palette_len = tga_palette_bits = + tga_x_origin = tga_y_origin = 0; + STBI_NOTUSED(tga_palette_start); + // OK, done + return tga_data; +} +#endif + +// ************************************************************************************************* +// Photoshop PSD loader -- PD by Thatcher Ulrich, integration by Nicolas Schulz, tweaked by STB + +#ifndef STBI_NO_PSD +static int stbi__psd_test(stbi__context *s) +{ + int r = (stbi__get32be(s) == 0x38425053); + stbi__rewind(s); + return r; +} + +static int stbi__psd_decode_rle(stbi__context *s, stbi_uc *p, int pixelCount) +{ + int count, nleft, len; + + count = 0; + while ((nleft = pixelCount - count) > 0) { + len = stbi__get8(s); + if (len == 128) { + // No-op. + } else if (len < 128) { + // Copy next len+1 bytes literally. + len++; + if (len > nleft) return 0; // corrupt data + count += len; + while (len) { + *p = stbi__get8(s); + p += 4; + len--; + } + } else if (len > 128) { + stbi_uc val; + // Next -len+1 bytes in the dest are replicated from next source byte. + // (Interpret len as a negative 8-bit int.) + len = 257 - len; + if (len > nleft) return 0; // corrupt data + val = stbi__get8(s); + count += len; + while (len) { + *p = val; + p += 4; + len--; + } + } + } + + return 1; +} + +static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc) +{ + int pixelCount; + int channelCount, compression; + int channel, i; + int bitdepth; + int w,h; + stbi_uc *out; + STBI_NOTUSED(ri); + + // Check identifier + if (stbi__get32be(s) != 0x38425053) // "8BPS" + return stbi__errpuc("not PSD", "Corrupt PSD image"); + + // Check file type version. + if (stbi__get16be(s) != 1) + return stbi__errpuc("wrong version", "Unsupported version of PSD image"); + + // Skip 6 reserved bytes. + stbi__skip(s, 6 ); + + // Read the number of channels (R, G, B, A, etc). + channelCount = stbi__get16be(s); + if (channelCount < 0 || channelCount > 16) + return stbi__errpuc("wrong channel count", "Unsupported number of channels in PSD image"); + + // Read the rows and columns of the image. + h = stbi__get32be(s); + w = stbi__get32be(s); + + if (h > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + if (w > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + + // Make sure the depth is 8 bits. + bitdepth = stbi__get16be(s); + if (bitdepth != 8 && bitdepth != 16) + return stbi__errpuc("unsupported bit depth", "PSD bit depth is not 8 or 16 bit"); + + // Make sure the color mode is RGB. + // Valid options are: + // 0: Bitmap + // 1: Grayscale + // 2: Indexed color + // 3: RGB color + // 4: CMYK color + // 7: Multichannel + // 8: Duotone + // 9: Lab color + if (stbi__get16be(s) != 3) + return stbi__errpuc("wrong color format", "PSD is not in RGB color format"); + + // Skip the Mode Data. (It's the palette for indexed color; other info for other modes.) + stbi__skip(s,stbi__get32be(s) ); + + // Skip the image resources. (resolution, pen tool paths, etc) + stbi__skip(s, stbi__get32be(s) ); + + // Skip the reserved data. + stbi__skip(s, stbi__get32be(s) ); + + // Find out if the data is compressed. + // Known values: + // 0: no compression + // 1: RLE compressed + compression = stbi__get16be(s); + if (compression > 1) + return stbi__errpuc("bad compression", "PSD has an unknown compression format"); + + // Check size + if (!stbi__mad3sizes_valid(4, w, h, 0)) + return stbi__errpuc("too large", "Corrupt PSD"); + + // Create the destination image. + + if (!compression && bitdepth == 16 && bpc == 16) { + out = (stbi_uc *) stbi__malloc_mad3(8, w, h, 0); + ri->bits_per_channel = 16; + } else + out = (stbi_uc *) stbi__malloc(4 * w*h); + + if (!out) return stbi__errpuc("outofmem", "Out of memory"); + pixelCount = w*h; + + // Initialize the data to zero. + //memset( out, 0, pixelCount * 4 ); + + // Finally, the image data. + if (compression) { + // RLE as used by .PSD and .TIFF + // Loop until you get the number of unpacked bytes you are expecting: + // Read the next source byte into n. + // If n is between 0 and 127 inclusive, copy the next n+1 bytes literally. + // Else if n is between -127 and -1 inclusive, copy the next byte -n+1 times. + // Else if n is 128, noop. + // Endloop + + // The RLE-compressed data is preceded by a 2-byte data count for each row in the data, + // which we're going to just skip. + stbi__skip(s, h * channelCount * 2 ); + + // Read the RLE data by channel. + for (channel = 0; channel < 4; channel++) { + stbi_uc *p; + + p = out+channel; + if (channel >= channelCount) { + // Fill this channel with default data. + for (i = 0; i < pixelCount; i++, p += 4) + *p = (channel == 3 ? 255 : 0); + } else { + // Read the RLE data. + if (!stbi__psd_decode_rle(s, p, pixelCount)) { + STBI_FREE(out); + return stbi__errpuc("corrupt", "bad RLE data"); + } + } + } + + } else { + // We're at the raw image data. It's each channel in order (Red, Green, Blue, Alpha, ...) + // where each channel consists of an 8-bit (or 16-bit) value for each pixel in the image. + + // Read the data by channel. + for (channel = 0; channel < 4; channel++) { + if (channel >= channelCount) { + // Fill this channel with default data. + if (bitdepth == 16 && bpc == 16) { + stbi__uint16 *q = ((stbi__uint16 *) out) + channel; + stbi__uint16 val = channel == 3 ? 65535 : 0; + for (i = 0; i < pixelCount; i++, q += 4) + *q = val; + } else { + stbi_uc *p = out+channel; + stbi_uc val = channel == 3 ? 255 : 0; + for (i = 0; i < pixelCount; i++, p += 4) + *p = val; + } + } else { + if (ri->bits_per_channel == 16) { // output bpc + stbi__uint16 *q = ((stbi__uint16 *) out) + channel; + for (i = 0; i < pixelCount; i++, q += 4) + *q = (stbi__uint16) stbi__get16be(s); + } else { + stbi_uc *p = out+channel; + if (bitdepth == 16) { // input bpc + for (i = 0; i < pixelCount; i++, p += 4) + *p = (stbi_uc) (stbi__get16be(s) >> 8); + } else { + for (i = 0; i < pixelCount; i++, p += 4) + *p = stbi__get8(s); + } + } + } + } + } + + // remove weird white matte from PSD + if (channelCount >= 4) { + if (ri->bits_per_channel == 16) { + for (i=0; i < w*h; ++i) { + stbi__uint16 *pixel = (stbi__uint16 *) out + 4*i; + if (pixel[3] != 0 && pixel[3] != 65535) { + float a = pixel[3] / 65535.0f; + float ra = 1.0f / a; + float inv_a = 65535.0f * (1 - ra); + pixel[0] = (stbi__uint16) (pixel[0]*ra + inv_a); + pixel[1] = (stbi__uint16) (pixel[1]*ra + inv_a); + pixel[2] = (stbi__uint16) (pixel[2]*ra + inv_a); + } + } + } else { + for (i=0; i < w*h; ++i) { + unsigned char *pixel = out + 4*i; + if (pixel[3] != 0 && pixel[3] != 255) { + float a = pixel[3] / 255.0f; + float ra = 1.0f / a; + float inv_a = 255.0f * (1 - ra); + pixel[0] = (unsigned char) (pixel[0]*ra + inv_a); + pixel[1] = (unsigned char) (pixel[1]*ra + inv_a); + pixel[2] = (unsigned char) (pixel[2]*ra + inv_a); + } + } + } + } + + // convert to desired output format + if (req_comp && req_comp != 4) { + if (ri->bits_per_channel == 16) + out = (stbi_uc *) stbi__convert_format16((stbi__uint16 *) out, 4, req_comp, w, h); + else + out = stbi__convert_format(out, 4, req_comp, w, h); + if (out == NULL) return out; // stbi__convert_format frees input on failure + } + + if (comp) *comp = 4; + *y = h; + *x = w; + + return out; +} +#endif + +// ************************************************************************************************* +// Softimage PIC loader +// by Tom Seddon +// +// See http://softimage.wiki.softimage.com/index.php/INFO:_PIC_file_format +// See http://ozviz.wasp.uwa.edu.au/~pbourke/dataformats/softimagepic/ + +#ifndef STBI_NO_PIC +static int stbi__pic_is4(stbi__context *s,const char *str) +{ + int i; + for (i=0; i<4; ++i) + if (stbi__get8(s) != (stbi_uc)str[i]) + return 0; + + return 1; +} + +static int stbi__pic_test_core(stbi__context *s) +{ + int i; + + if (!stbi__pic_is4(s,"\x53\x80\xF6\x34")) + return 0; + + for(i=0;i<84;++i) + stbi__get8(s); + + if (!stbi__pic_is4(s,"PICT")) + return 0; + + return 1; +} + +typedef struct +{ + stbi_uc size,type,channel; +} stbi__pic_packet; + +static stbi_uc *stbi__readval(stbi__context *s, int channel, stbi_uc *dest) +{ + int mask=0x80, i; + + for (i=0; i<4; ++i, mask>>=1) { + if (channel & mask) { + if (stbi__at_eof(s)) return stbi__errpuc("bad file","PIC file too short"); + dest[i]=stbi__get8(s); + } + } + + return dest; +} + +static void stbi__copyval(int channel,stbi_uc *dest,const stbi_uc *src) +{ + int mask=0x80,i; + + for (i=0;i<4; ++i, mask>>=1) + if (channel&mask) + dest[i]=src[i]; +} + +static stbi_uc *stbi__pic_load_core(stbi__context *s,int width,int height,int *comp, stbi_uc *result) +{ + int act_comp=0,num_packets=0,y,chained; + stbi__pic_packet packets[10]; + + // this will (should...) cater for even some bizarre stuff like having data + // for the same channel in multiple packets. + do { + stbi__pic_packet *packet; + + if (num_packets==sizeof(packets)/sizeof(packets[0])) + return stbi__errpuc("bad format","too many packets"); + + packet = &packets[num_packets++]; + + chained = stbi__get8(s); + packet->size = stbi__get8(s); + packet->type = stbi__get8(s); + packet->channel = stbi__get8(s); + + act_comp |= packet->channel; + + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (reading packets)"); + if (packet->size != 8) return stbi__errpuc("bad format","packet isn't 8bpp"); + } while (chained); + + *comp = (act_comp & 0x10 ? 4 : 3); // has alpha channel? + + for(y=0; ytype) { + default: + return stbi__errpuc("bad format","packet has bad compression type"); + + case 0: {//uncompressed + int x; + + for(x=0;xchannel,dest)) + return 0; + break; + } + + case 1://Pure RLE + { + int left=width, i; + + while (left>0) { + stbi_uc count,value[4]; + + count=stbi__get8(s); + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (pure read count)"); + + if (count > left) + count = (stbi_uc) left; + + if (!stbi__readval(s,packet->channel,value)) return 0; + + for(i=0; ichannel,dest,value); + left -= count; + } + } + break; + + case 2: {//Mixed RLE + int left=width; + while (left>0) { + int count = stbi__get8(s), i; + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (mixed read count)"); + + if (count >= 128) { // Repeated + stbi_uc value[4]; + + if (count==128) + count = stbi__get16be(s); + else + count -= 127; + if (count > left) + return stbi__errpuc("bad file","scanline overrun"); + + if (!stbi__readval(s,packet->channel,value)) + return 0; + + for(i=0;ichannel,dest,value); + } else { // Raw + ++count; + if (count>left) return stbi__errpuc("bad file","scanline overrun"); + + for(i=0;ichannel,dest)) + return 0; + } + left-=count; + } + break; + } + } + } + } + + return result; +} + +static void *stbi__pic_load(stbi__context *s,int *px,int *py,int *comp,int req_comp, stbi__result_info *ri) +{ + stbi_uc *result; + int i, x,y, internal_comp; + STBI_NOTUSED(ri); + + if (!comp) comp = &internal_comp; + + for (i=0; i<92; ++i) + stbi__get8(s); + + x = stbi__get16be(s); + y = stbi__get16be(s); + + if (y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + if (x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (pic header)"); + if (!stbi__mad3sizes_valid(x, y, 4, 0)) return stbi__errpuc("too large", "PIC image too large to decode"); + + stbi__get32be(s); //skip `ratio' + stbi__get16be(s); //skip `fields' + stbi__get16be(s); //skip `pad' + + // intermediate buffer is RGBA + result = (stbi_uc *) stbi__malloc_mad3(x, y, 4, 0); + if (!result) return stbi__errpuc("outofmem", "Out of memory"); + memset(result, 0xff, x*y*4); + + if (!stbi__pic_load_core(s,x,y,comp, result)) { + STBI_FREE(result); + result=0; + } + *px = x; + *py = y; + if (req_comp == 0) req_comp = *comp; + result=stbi__convert_format(result,4,req_comp,x,y); + + return result; +} + +static int stbi__pic_test(stbi__context *s) +{ + int r = stbi__pic_test_core(s); + stbi__rewind(s); + return r; +} +#endif + +// ************************************************************************************************* +// GIF loader -- public domain by Jean-Marc Lienher -- simplified/shrunk by stb + +#ifndef STBI_NO_GIF +typedef struct +{ + stbi__int16 prefix; + stbi_uc first; + stbi_uc suffix; +} stbi__gif_lzw; + +typedef struct +{ + int w,h; + stbi_uc *out; // output buffer (always 4 components) + stbi_uc *background; // The current "background" as far as a gif is concerned + stbi_uc *history; + int flags, bgindex, ratio, transparent, eflags; + stbi_uc pal[256][4]; + stbi_uc lpal[256][4]; + stbi__gif_lzw codes[8192]; + stbi_uc *color_table; + int parse, step; + int lflags; + int start_x, start_y; + int max_x, max_y; + int cur_x, cur_y; + int line_size; + int delay; +} stbi__gif; + +static int stbi__gif_test_raw(stbi__context *s) +{ + int sz; + if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') return 0; + sz = stbi__get8(s); + if (sz != '9' && sz != '7') return 0; + if (stbi__get8(s) != 'a') return 0; + return 1; +} + +static int stbi__gif_test(stbi__context *s) +{ + int r = stbi__gif_test_raw(s); + stbi__rewind(s); + return r; +} + +static void stbi__gif_parse_colortable(stbi__context *s, stbi_uc pal[256][4], int num_entries, int transp) +{ + int i; + for (i=0; i < num_entries; ++i) { + pal[i][2] = stbi__get8(s); + pal[i][1] = stbi__get8(s); + pal[i][0] = stbi__get8(s); + pal[i][3] = transp == i ? 0 : 255; + } +} + +static int stbi__gif_header(stbi__context *s, stbi__gif *g, int *comp, int is_info) +{ + stbi_uc version; + if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') + return stbi__err("not GIF", "Corrupt GIF"); + + version = stbi__get8(s); + if (version != '7' && version != '9') return stbi__err("not GIF", "Corrupt GIF"); + if (stbi__get8(s) != 'a') return stbi__err("not GIF", "Corrupt GIF"); + + stbi__g_failure_reason = ""; + g->w = stbi__get16le(s); + g->h = stbi__get16le(s); + g->flags = stbi__get8(s); + g->bgindex = stbi__get8(s); + g->ratio = stbi__get8(s); + g->transparent = -1; + + if (g->w > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)"); + if (g->h > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)"); + + if (comp != 0) *comp = 4; // can't actually tell whether it's 3 or 4 until we parse the comments + + if (is_info) return 1; + + if (g->flags & 0x80) + stbi__gif_parse_colortable(s,g->pal, 2 << (g->flags & 7), -1); + + return 1; +} + +static int stbi__gif_info_raw(stbi__context *s, int *x, int *y, int *comp) +{ + stbi__gif* g = (stbi__gif*) stbi__malloc(sizeof(stbi__gif)); + if (!g) return stbi__err("outofmem", "Out of memory"); + if (!stbi__gif_header(s, g, comp, 1)) { + STBI_FREE(g); + stbi__rewind( s ); + return 0; + } + if (x) *x = g->w; + if (y) *y = g->h; + STBI_FREE(g); + return 1; +} + +static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code) +{ + stbi_uc *p, *c; + int idx; + + // recurse to decode the prefixes, since the linked-list is backwards, + // and working backwards through an interleaved image would be nasty + if (g->codes[code].prefix >= 0) + stbi__out_gif_code(g, g->codes[code].prefix); + + if (g->cur_y >= g->max_y) return; + + idx = g->cur_x + g->cur_y; + p = &g->out[idx]; + g->history[idx / 4] = 1; + + c = &g->color_table[g->codes[code].suffix * 4]; + if (c[3] > 128) { // don't render transparent pixels; + p[0] = c[2]; + p[1] = c[1]; + p[2] = c[0]; + p[3] = c[3]; + } + g->cur_x += 4; + + if (g->cur_x >= g->max_x) { + g->cur_x = g->start_x; + g->cur_y += g->step; + + while (g->cur_y >= g->max_y && g->parse > 0) { + g->step = (1 << g->parse) * g->line_size; + g->cur_y = g->start_y + (g->step >> 1); + --g->parse; + } + } +} + +static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g) +{ + stbi_uc lzw_cs; + stbi__int32 len, init_code; + stbi__uint32 first; + stbi__int32 codesize, codemask, avail, oldcode, bits, valid_bits, clear; + stbi__gif_lzw *p; + + lzw_cs = stbi__get8(s); + if (lzw_cs > 12) return NULL; + clear = 1 << lzw_cs; + first = 1; + codesize = lzw_cs + 1; + codemask = (1 << codesize) - 1; + bits = 0; + valid_bits = 0; + for (init_code = 0; init_code < clear; init_code++) { + g->codes[init_code].prefix = -1; + g->codes[init_code].first = (stbi_uc) init_code; + g->codes[init_code].suffix = (stbi_uc) init_code; + } + + // support no starting clear code + avail = clear+2; + oldcode = -1; + + len = 0; + for(;;) { + if (valid_bits < codesize) { + if (len == 0) { + len = stbi__get8(s); // start new block + if (len == 0) + return g->out; + } + --len; + bits |= (stbi__int32) stbi__get8(s) << valid_bits; + valid_bits += 8; + } else { + stbi__int32 code = bits & codemask; + bits >>= codesize; + valid_bits -= codesize; + // @OPTIMIZE: is there some way we can accelerate the non-clear path? + if (code == clear) { // clear code + codesize = lzw_cs + 1; + codemask = (1 << codesize) - 1; + avail = clear + 2; + oldcode = -1; + first = 0; + } else if (code == clear + 1) { // end of stream code + stbi__skip(s, len); + while ((len = stbi__get8(s)) > 0) + stbi__skip(s,len); + return g->out; + } else if (code <= avail) { + if (first) { + return stbi__errpuc("no clear code", "Corrupt GIF"); + } + + if (oldcode >= 0) { + p = &g->codes[avail++]; + if (avail > 8192) { + return stbi__errpuc("too many codes", "Corrupt GIF"); + } + + p->prefix = (stbi__int16) oldcode; + p->first = g->codes[oldcode].first; + p->suffix = (code == avail) ? p->first : g->codes[code].first; + } else if (code == avail) + return stbi__errpuc("illegal code in raster", "Corrupt GIF"); + + stbi__out_gif_code(g, (stbi__uint16) code); + + if ((avail & codemask) == 0 && avail <= 0x0FFF) { + codesize++; + codemask = (1 << codesize) - 1; + } + + oldcode = code; + } else { + return stbi__errpuc("illegal code in raster", "Corrupt GIF"); + } + } + } +} + +// this function is designed to support animated gifs, although stb_image doesn't support it +// two back is the image from two frames ago, used for a very specific disposal format +static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, int req_comp, stbi_uc *two_back) +{ + int dispose; + int first_frame; + int pi; + int pcount; + STBI_NOTUSED(req_comp); + + // on first frame, any non-written pixels get the background colour (non-transparent) + first_frame = 0; + if (g->out == 0) { + if (!stbi__gif_header(s, g, comp,0)) return 0; // stbi__g_failure_reason set by stbi__gif_header + if (!stbi__mad3sizes_valid(4, g->w, g->h, 0)) + return stbi__errpuc("too large", "GIF image is too large"); + pcount = g->w * g->h; + g->out = (stbi_uc *) stbi__malloc(4 * pcount); + g->background = (stbi_uc *) stbi__malloc(4 * pcount); + g->history = (stbi_uc *) stbi__malloc(pcount); + if (!g->out || !g->background || !g->history) + return stbi__errpuc("outofmem", "Out of memory"); + + // image is treated as "transparent" at the start - ie, nothing overwrites the current background; + // background colour is only used for pixels that are not rendered first frame, after that "background" + // color refers to the color that was there the previous frame. + memset(g->out, 0x00, 4 * pcount); + memset(g->background, 0x00, 4 * pcount); // state of the background (starts transparent) + memset(g->history, 0x00, pcount); // pixels that were affected previous frame + first_frame = 1; + } else { + // second frame - how do we dispose of the previous one? + dispose = (g->eflags & 0x1C) >> 2; + pcount = g->w * g->h; + + if ((dispose == 3) && (two_back == 0)) { + dispose = 2; // if I don't have an image to revert back to, default to the old background + } + + if (dispose == 3) { // use previous graphic + for (pi = 0; pi < pcount; ++pi) { + if (g->history[pi]) { + memcpy( &g->out[pi * 4], &two_back[pi * 4], 4 ); + } + } + } else if (dispose == 2) { + // restore what was changed last frame to background before that frame; + for (pi = 0; pi < pcount; ++pi) { + if (g->history[pi]) { + memcpy( &g->out[pi * 4], &g->background[pi * 4], 4 ); + } + } + } else { + // This is a non-disposal case eithe way, so just + // leave the pixels as is, and they will become the new background + // 1: do not dispose + // 0: not specified. + } + + // background is what out is after the undoing of the previou frame; + memcpy( g->background, g->out, 4 * g->w * g->h ); + } + + // clear my history; + memset( g->history, 0x00, g->w * g->h ); // pixels that were affected previous frame + + for (;;) { + int tag = stbi__get8(s); + switch (tag) { + case 0x2C: /* Image Descriptor */ + { + stbi__int32 x, y, w, h; + stbi_uc *o; + + x = stbi__get16le(s); + y = stbi__get16le(s); + w = stbi__get16le(s); + h = stbi__get16le(s); + if (((x + w) > (g->w)) || ((y + h) > (g->h))) + return stbi__errpuc("bad Image Descriptor", "Corrupt GIF"); + + g->line_size = g->w * 4; + g->start_x = x * 4; + g->start_y = y * g->line_size; + g->max_x = g->start_x + w * 4; + g->max_y = g->start_y + h * g->line_size; + g->cur_x = g->start_x; + g->cur_y = g->start_y; + + // if the width of the specified rectangle is 0, that means + // we may not see *any* pixels or the image is malformed; + // to make sure this is caught, move the current y down to + // max_y (which is what out_gif_code checks). + if (w == 0) + g->cur_y = g->max_y; + + g->lflags = stbi__get8(s); + + if (g->lflags & 0x40) { + g->step = 8 * g->line_size; // first interlaced spacing + g->parse = 3; + } else { + g->step = g->line_size; + g->parse = 0; + } + + if (g->lflags & 0x80) { + stbi__gif_parse_colortable(s,g->lpal, 2 << (g->lflags & 7), g->eflags & 0x01 ? g->transparent : -1); + g->color_table = (stbi_uc *) g->lpal; + } else if (g->flags & 0x80) { + g->color_table = (stbi_uc *) g->pal; + } else + return stbi__errpuc("missing color table", "Corrupt GIF"); + + o = stbi__process_gif_raster(s, g); + if (!o) return NULL; + + // if this was the first frame, + pcount = g->w * g->h; + if (first_frame && (g->bgindex > 0)) { + // if first frame, any pixel not drawn to gets the background color + for (pi = 0; pi < pcount; ++pi) { + if (g->history[pi] == 0) { + g->pal[g->bgindex][3] = 255; // just in case it was made transparent, undo that; It will be reset next frame if need be; + memcpy( &g->out[pi * 4], &g->pal[g->bgindex], 4 ); + } + } + } + + return o; + } + + case 0x21: // Comment Extension. + { + int len; + int ext = stbi__get8(s); + if (ext == 0xF9) { // Graphic Control Extension. + len = stbi__get8(s); + if (len == 4) { + g->eflags = stbi__get8(s); + g->delay = 10 * stbi__get16le(s); // delay - 1/100th of a second, saving as 1/1000ths. + + // unset old transparent + if (g->transparent >= 0) { + g->pal[g->transparent][3] = 255; + } + if (g->eflags & 0x01) { + g->transparent = stbi__get8(s); + if (g->transparent >= 0) { + g->pal[g->transparent][3] = 0; + } + } else { + // don't need transparent + stbi__skip(s, 1); + g->transparent = -1; + } + } else { + stbi__skip(s, len); + break; + } + } + while ((len = stbi__get8(s)) != 0) { + stbi__skip(s, len); + } + break; + } + + case 0x3B: // gif stream termination code + return (stbi_uc *) s; // using '1' causes warning on some compilers + + default: + return stbi__errpuc("unknown code", "Corrupt GIF"); + } + } +} + +static void *stbi__load_gif_main_outofmem(stbi__gif *g, stbi_uc *out, int **delays) +{ + STBI_FREE(g->out); + STBI_FREE(g->history); + STBI_FREE(g->background); + + if (out) STBI_FREE(out); + if (delays && *delays) STBI_FREE(*delays); + return stbi__errpuc("outofmem", "Out of memory"); +} + +static void *stbi__load_gif_main(stbi__context *s, int **delays, int *x, int *y, int *z, int *comp, int req_comp) +{ + if (stbi__gif_test(s)) { + int layers = 0; + stbi_uc *u = 0; + stbi_uc *out = 0; + stbi_uc *two_back = 0; + stbi__gif g; + int stride; + int out_size = 0; + int delays_size = 0; + + STBI_NOTUSED(out_size); + STBI_NOTUSED(delays_size); + + memset(&g, 0, sizeof(g)); + if (delays) { + *delays = 0; + } + + do { + u = stbi__gif_load_next(s, &g, comp, req_comp, two_back); + if (u == (stbi_uc *) s) u = 0; // end of animated gif marker + + if (u) { + *x = g.w; + *y = g.h; + ++layers; + stride = g.w * g.h * 4; + + if (out) { + void *tmp = (stbi_uc*) STBI_REALLOC_SIZED( out, out_size, layers * stride ); + if (!tmp) + return stbi__load_gif_main_outofmem(&g, out, delays); + else { + out = (stbi_uc*) tmp; + out_size = layers * stride; + } + + if (delays) { + int *new_delays = (int*) STBI_REALLOC_SIZED( *delays, delays_size, sizeof(int) * layers ); + if (!new_delays) + return stbi__load_gif_main_outofmem(&g, out, delays); + *delays = new_delays; + delays_size = layers * sizeof(int); + } + } else { + out = (stbi_uc*)stbi__malloc( layers * stride ); + if (!out) + return stbi__load_gif_main_outofmem(&g, out, delays); + out_size = layers * stride; + if (delays) { + *delays = (int*) stbi__malloc( layers * sizeof(int) ); + if (!*delays) + return stbi__load_gif_main_outofmem(&g, out, delays); + delays_size = layers * sizeof(int); + } + } + memcpy( out + ((layers - 1) * stride), u, stride ); + if (layers >= 2) { + two_back = out - 2 * stride; + } + + if (delays) { + (*delays)[layers - 1U] = g.delay; + } + } + } while (u != 0); + + // free temp buffer; + STBI_FREE(g.out); + STBI_FREE(g.history); + STBI_FREE(g.background); + + // do the final conversion after loading everything; + if (req_comp && req_comp != 4) + out = stbi__convert_format(out, 4, req_comp, layers * g.w, g.h); + + *z = layers; + return out; + } else { + return stbi__errpuc("not GIF", "Image was not as a gif type."); + } +} + +static void *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri) +{ + stbi_uc *u = 0; + stbi__gif g; + memset(&g, 0, sizeof(g)); + STBI_NOTUSED(ri); + + u = stbi__gif_load_next(s, &g, comp, req_comp, 0); + if (u == (stbi_uc *) s) u = 0; // end of animated gif marker + if (u) { + *x = g.w; + *y = g.h; + + // moved conversion to after successful load so that the same + // can be done for multiple frames. + if (req_comp && req_comp != 4) + u = stbi__convert_format(u, 4, req_comp, g.w, g.h); + } else if (g.out) { + // if there was an error and we allocated an image buffer, free it! + STBI_FREE(g.out); + } + + // free buffers needed for multiple frame loading; + STBI_FREE(g.history); + STBI_FREE(g.background); + + return u; +} + +static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp) +{ + return stbi__gif_info_raw(s,x,y,comp); +} +#endif + +// ************************************************************************************************* +// Radiance RGBE HDR loader +// originally by Nicolas Schulz +#ifndef STBI_NO_HDR +static int stbi__hdr_test_core(stbi__context *s, const char *signature) +{ + int i; + for (i=0; signature[i]; ++i) + if (stbi__get8(s) != signature[i]) + return 0; + stbi__rewind(s); + return 1; +} + +static int stbi__hdr_test(stbi__context* s) +{ + int r = stbi__hdr_test_core(s, "#?RADIANCE\n"); + stbi__rewind(s); + if(!r) { + r = stbi__hdr_test_core(s, "#?RGBE\n"); + stbi__rewind(s); + } + return r; +} + +#define STBI__HDR_BUFLEN 1024 +static char *stbi__hdr_gettoken(stbi__context *z, char *buffer) +{ + int len=0; + char c = '\0'; + + c = (char) stbi__get8(z); + + while (!stbi__at_eof(z) && c != '\n') { + buffer[len++] = c; + if (len == STBI__HDR_BUFLEN-1) { + // flush to end of line + while (!stbi__at_eof(z) && stbi__get8(z) != '\n') + ; + break; + } + c = (char) stbi__get8(z); + } + + buffer[len] = 0; + return buffer; +} + +static void stbi__hdr_convert(float *output, stbi_uc *input, int req_comp) +{ + if ( input[3] != 0 ) { + float f1; + // Exponent + f1 = (float) ldexp(1.0f, input[3] - (int)(128 + 8)); + if (req_comp <= 2) + output[0] = (input[0] + input[1] + input[2]) * f1 / 3; + else { + output[0] = input[0] * f1; + output[1] = input[1] * f1; + output[2] = input[2] * f1; + } + if (req_comp == 2) output[1] = 1; + if (req_comp == 4) output[3] = 1; + } else { + switch (req_comp) { + case 4: output[3] = 1; /* fallthrough */ + case 3: output[0] = output[1] = output[2] = 0; + break; + case 2: output[1] = 1; /* fallthrough */ + case 1: output[0] = 0; + break; + } + } +} + +static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri) +{ + char buffer[STBI__HDR_BUFLEN]; + char *token; + int valid = 0; + int width, height; + stbi_uc *scanline; + float *hdr_data; + int len; + unsigned char count, value; + int i, j, k, c1,c2, z; + const char *headerToken; + STBI_NOTUSED(ri); + + // Check identifier + headerToken = stbi__hdr_gettoken(s,buffer); + if (strcmp(headerToken, "#?RADIANCE") != 0 && strcmp(headerToken, "#?RGBE") != 0) + return stbi__errpf("not HDR", "Corrupt HDR image"); + + // Parse header + for(;;) { + token = stbi__hdr_gettoken(s,buffer); + if (token[0] == 0) break; + if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1; + } + + if (!valid) return stbi__errpf("unsupported format", "Unsupported HDR format"); + + // Parse width and height + // can't use sscanf() if we're not using stdio! + token = stbi__hdr_gettoken(s,buffer); + if (strncmp(token, "-Y ", 3)) return stbi__errpf("unsupported data layout", "Unsupported HDR format"); + token += 3; + height = (int) strtol(token, &token, 10); + while (*token == ' ') ++token; + if (strncmp(token, "+X ", 3)) return stbi__errpf("unsupported data layout", "Unsupported HDR format"); + token += 3; + width = (int) strtol(token, NULL, 10); + + if (height > STBI_MAX_DIMENSIONS) return stbi__errpf("too large","Very large image (corrupt?)"); + if (width > STBI_MAX_DIMENSIONS) return stbi__errpf("too large","Very large image (corrupt?)"); + + *x = width; + *y = height; + + if (comp) *comp = 3; + if (req_comp == 0) req_comp = 3; + + if (!stbi__mad4sizes_valid(width, height, req_comp, sizeof(float), 0)) + return stbi__errpf("too large", "HDR image is too large"); + + // Read data + hdr_data = (float *) stbi__malloc_mad4(width, height, req_comp, sizeof(float), 0); + if (!hdr_data) + return stbi__errpf("outofmem", "Out of memory"); + + // Load image data + // image data is stored as some number of sca + if ( width < 8 || width >= 32768) { + // Read flat data + for (j=0; j < height; ++j) { + for (i=0; i < width; ++i) { + stbi_uc rgbe[4]; + main_decode_loop: + stbi__getn(s, rgbe, 4); + stbi__hdr_convert(hdr_data + j * width * req_comp + i * req_comp, rgbe, req_comp); + } + } + } else { + // Read RLE-encoded data + scanline = NULL; + + for (j = 0; j < height; ++j) { + c1 = stbi__get8(s); + c2 = stbi__get8(s); + len = stbi__get8(s); + if (c1 != 2 || c2 != 2 || (len & 0x80)) { + // not run-length encoded, so we have to actually use THIS data as a decoded + // pixel (note this can't be a valid pixel--one of RGB must be >= 128) + stbi_uc rgbe[4]; + rgbe[0] = (stbi_uc) c1; + rgbe[1] = (stbi_uc) c2; + rgbe[2] = (stbi_uc) len; + rgbe[3] = (stbi_uc) stbi__get8(s); + stbi__hdr_convert(hdr_data, rgbe, req_comp); + i = 1; + j = 0; + STBI_FREE(scanline); + goto main_decode_loop; // yes, this makes no sense + } + len <<= 8; + len |= stbi__get8(s); + if (len != width) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("invalid decoded scanline length", "corrupt HDR"); } + if (scanline == NULL) { + scanline = (stbi_uc *) stbi__malloc_mad2(width, 4, 0); + if (!scanline) { + STBI_FREE(hdr_data); + return stbi__errpf("outofmem", "Out of memory"); + } + } + + for (k = 0; k < 4; ++k) { + int nleft; + i = 0; + while ((nleft = width - i) > 0) { + count = stbi__get8(s); + if (count > 128) { + // Run + value = stbi__get8(s); + count -= 128; + if ((count == 0) || (count > nleft)) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("corrupt", "bad RLE data in HDR"); } + for (z = 0; z < count; ++z) + scanline[i++ * 4 + k] = value; + } else { + // Dump + if ((count == 0) || (count > nleft)) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("corrupt", "bad RLE data in HDR"); } + for (z = 0; z < count; ++z) + scanline[i++ * 4 + k] = stbi__get8(s); + } + } + } + for (i=0; i < width; ++i) + stbi__hdr_convert(hdr_data+(j*width + i)*req_comp, scanline + i*4, req_comp); + } + if (scanline) + STBI_FREE(scanline); + } + + return hdr_data; +} + +static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp) +{ + char buffer[STBI__HDR_BUFLEN]; + char *token; + int valid = 0; + int dummy; + + if (!x) x = &dummy; + if (!y) y = &dummy; + if (!comp) comp = &dummy; + + if (stbi__hdr_test(s) == 0) { + stbi__rewind( s ); + return 0; + } + + for(;;) { + token = stbi__hdr_gettoken(s,buffer); + if (token[0] == 0) break; + if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1; + } + + if (!valid) { + stbi__rewind( s ); + return 0; + } + token = stbi__hdr_gettoken(s,buffer); + if (strncmp(token, "-Y ", 3)) { + stbi__rewind( s ); + return 0; + } + token += 3; + *y = (int) strtol(token, &token, 10); + while (*token == ' ') ++token; + if (strncmp(token, "+X ", 3)) { + stbi__rewind( s ); + return 0; + } + token += 3; + *x = (int) strtol(token, NULL, 10); + *comp = 3; + return 1; +} +#endif // STBI_NO_HDR + +#ifndef STBI_NO_BMP +static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp) +{ + void *p; + stbi__bmp_data info; + + info.all_a = 255; + p = stbi__bmp_parse_header(s, &info); + if (p == NULL) { + stbi__rewind( s ); + return 0; + } + if (x) *x = s->img_x; + if (y) *y = s->img_y; + if (comp) { + if (info.bpp == 24 && info.ma == 0xff000000) + *comp = 3; + else + *comp = info.ma ? 4 : 3; + } + return 1; +} +#endif + +#ifndef STBI_NO_PSD +static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp) +{ + int channelCount, dummy, depth; + if (!x) x = &dummy; + if (!y) y = &dummy; + if (!comp) comp = &dummy; + if (stbi__get32be(s) != 0x38425053) { + stbi__rewind( s ); + return 0; + } + if (stbi__get16be(s) != 1) { + stbi__rewind( s ); + return 0; + } + stbi__skip(s, 6); + channelCount = stbi__get16be(s); + if (channelCount < 0 || channelCount > 16) { + stbi__rewind( s ); + return 0; + } + *y = stbi__get32be(s); + *x = stbi__get32be(s); + depth = stbi__get16be(s); + if (depth != 8 && depth != 16) { + stbi__rewind( s ); + return 0; + } + if (stbi__get16be(s) != 3) { + stbi__rewind( s ); + return 0; + } + *comp = 4; + return 1; +} + +static int stbi__psd_is16(stbi__context *s) +{ + int channelCount, depth; + if (stbi__get32be(s) != 0x38425053) { + stbi__rewind( s ); + return 0; + } + if (stbi__get16be(s) != 1) { + stbi__rewind( s ); + return 0; + } + stbi__skip(s, 6); + channelCount = stbi__get16be(s); + if (channelCount < 0 || channelCount > 16) { + stbi__rewind( s ); + return 0; + } + STBI_NOTUSED(stbi__get32be(s)); + STBI_NOTUSED(stbi__get32be(s)); + depth = stbi__get16be(s); + if (depth != 16) { + stbi__rewind( s ); + return 0; + } + return 1; +} +#endif + +#ifndef STBI_NO_PIC +static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp) +{ + int act_comp=0,num_packets=0,chained,dummy; + stbi__pic_packet packets[10]; + + if (!x) x = &dummy; + if (!y) y = &dummy; + if (!comp) comp = &dummy; + + if (!stbi__pic_is4(s,"\x53\x80\xF6\x34")) { + stbi__rewind(s); + return 0; + } + + stbi__skip(s, 88); + + *x = stbi__get16be(s); + *y = stbi__get16be(s); + if (stbi__at_eof(s)) { + stbi__rewind( s); + return 0; + } + if ( (*x) != 0 && (1 << 28) / (*x) < (*y)) { + stbi__rewind( s ); + return 0; + } + + stbi__skip(s, 8); + + do { + stbi__pic_packet *packet; + + if (num_packets==sizeof(packets)/sizeof(packets[0])) + return 0; + + packet = &packets[num_packets++]; + chained = stbi__get8(s); + packet->size = stbi__get8(s); + packet->type = stbi__get8(s); + packet->channel = stbi__get8(s); + act_comp |= packet->channel; + + if (stbi__at_eof(s)) { + stbi__rewind( s ); + return 0; + } + if (packet->size != 8) { + stbi__rewind( s ); + return 0; + } + } while (chained); + + *comp = (act_comp & 0x10 ? 4 : 3); + + return 1; +} +#endif + +// ************************************************************************************************* +// Portable Gray Map and Portable Pixel Map loader +// by Ken Miller +// +// PGM: http://netpbm.sourceforge.net/doc/pgm.html +// PPM: http://netpbm.sourceforge.net/doc/ppm.html +// +// Known limitations: +// Does not support comments in the header section +// Does not support ASCII image data (formats P2 and P3) + +#ifndef STBI_NO_PNM + +static int stbi__pnm_test(stbi__context *s) +{ + char p, t; + p = (char) stbi__get8(s); + t = (char) stbi__get8(s); + if (p != 'P' || (t != '5' && t != '6')) { + stbi__rewind( s ); + return 0; + } + return 1; +} + +static void *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri) +{ + stbi_uc *out; + STBI_NOTUSED(ri); + + ri->bits_per_channel = stbi__pnm_info(s, (int *)&s->img_x, (int *)&s->img_y, (int *)&s->img_n); + if (ri->bits_per_channel == 0) + return 0; + + if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)"); + + *x = s->img_x; + *y = s->img_y; + if (comp) *comp = s->img_n; + + if (!stbi__mad4sizes_valid(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0)) + return stbi__errpuc("too large", "PNM too large"); + + out = (stbi_uc *) stbi__malloc_mad4(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0); + if (!out) return stbi__errpuc("outofmem", "Out of memory"); + if (!stbi__getn(s, out, s->img_n * s->img_x * s->img_y * (ri->bits_per_channel / 8))) { + STBI_FREE(out); + return stbi__errpuc("bad PNM", "PNM file truncated"); + } + + if (req_comp && req_comp != s->img_n) { + if (ri->bits_per_channel == 16) { + out = (stbi_uc *) stbi__convert_format16((stbi__uint16 *) out, s->img_n, req_comp, s->img_x, s->img_y); + } else { + out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y); + } + if (out == NULL) return out; // stbi__convert_format frees input on failure + } + return out; +} + +static int stbi__pnm_isspace(char c) +{ + return c == ' ' || c == '\t' || c == '\n' || c == '\v' || c == '\f' || c == '\r'; +} + +static void stbi__pnm_skip_whitespace(stbi__context *s, char *c) +{ + for (;;) { + while (!stbi__at_eof(s) && stbi__pnm_isspace(*c)) + *c = (char) stbi__get8(s); + + if (stbi__at_eof(s) || *c != '#') + break; + + while (!stbi__at_eof(s) && *c != '\n' && *c != '\r' ) + *c = (char) stbi__get8(s); + } +} + +static int stbi__pnm_isdigit(char c) +{ + return c >= '0' && c <= '9'; +} + +static int stbi__pnm_getinteger(stbi__context *s, char *c) +{ + int value = 0; + + while (!stbi__at_eof(s) && stbi__pnm_isdigit(*c)) { + value = value*10 + (*c - '0'); + *c = (char) stbi__get8(s); + if((value > 214748364) || (value == 214748364 && *c > '7')) + return stbi__err("integer parse overflow", "Parsing an integer in the PPM header overflowed a 32-bit int"); + } + + return value; +} + +static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp) +{ + int maxv, dummy; + char c, p, t; + + if (!x) x = &dummy; + if (!y) y = &dummy; + if (!comp) comp = &dummy; + + stbi__rewind(s); + + // Get identifier + p = (char) stbi__get8(s); + t = (char) stbi__get8(s); + if (p != 'P' || (t != '5' && t != '6')) { + stbi__rewind(s); + return 0; + } + + *comp = (t == '6') ? 3 : 1; // '5' is 1-component .pgm; '6' is 3-component .ppm + + c = (char) stbi__get8(s); + stbi__pnm_skip_whitespace(s, &c); + + *x = stbi__pnm_getinteger(s, &c); // read width + if(*x == 0) + return stbi__err("invalid width", "PPM image header had zero or overflowing width"); + stbi__pnm_skip_whitespace(s, &c); + + *y = stbi__pnm_getinteger(s, &c); // read height + if (*y == 0) + return stbi__err("invalid width", "PPM image header had zero or overflowing width"); + stbi__pnm_skip_whitespace(s, &c); + + maxv = stbi__pnm_getinteger(s, &c); // read max value + if (maxv > 65535) + return stbi__err("max value > 65535", "PPM image supports only 8-bit and 16-bit images"); + else if (maxv > 255) + return 16; + else + return 8; +} + +static int stbi__pnm_is16(stbi__context *s) +{ + if (stbi__pnm_info(s, NULL, NULL, NULL) == 16) + return 1; + return 0; +} +#endif + +static int stbi__info_main(stbi__context *s, int *x, int *y, int *comp) +{ + #ifndef STBI_NO_JPEG + if (stbi__jpeg_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PNG + if (stbi__png_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_GIF + if (stbi__gif_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_BMP + if (stbi__bmp_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PSD + if (stbi__psd_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PIC + if (stbi__pic_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PNM + if (stbi__pnm_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_HDR + if (stbi__hdr_info(s, x, y, comp)) return 1; + #endif + + // test tga last because it's a crappy test! + #ifndef STBI_NO_TGA + if (stbi__tga_info(s, x, y, comp)) + return 1; + #endif + return stbi__err("unknown image type", "Image not of any known type, or corrupt"); +} + +static int stbi__is_16_main(stbi__context *s) +{ + #ifndef STBI_NO_PNG + if (stbi__png_is16(s)) return 1; + #endif + + #ifndef STBI_NO_PSD + if (stbi__psd_is16(s)) return 1; + #endif + + #ifndef STBI_NO_PNM + if (stbi__pnm_is16(s)) return 1; + #endif + return 0; +} + +#ifndef STBI_NO_STDIO +STBIDEF int stbi_info(char const *filename, int *x, int *y, int *comp) +{ + FILE *f = stbi__fopen(filename, "rb"); + int result; + if (!f) return stbi__err("can't fopen", "Unable to open file"); + result = stbi_info_from_file(f, x, y, comp); + fclose(f); + return result; +} + +STBIDEF int stbi_info_from_file(FILE *f, int *x, int *y, int *comp) +{ + int r; + stbi__context s; + long pos = ftell(f); + stbi__start_file(&s, f); + r = stbi__info_main(&s,x,y,comp); + fseek(f,pos,SEEK_SET); + return r; +} + +STBIDEF int stbi_is_16_bit(char const *filename) +{ + FILE *f = stbi__fopen(filename, "rb"); + int result; + if (!f) return stbi__err("can't fopen", "Unable to open file"); + result = stbi_is_16_bit_from_file(f); + fclose(f); + return result; +} + +STBIDEF int stbi_is_16_bit_from_file(FILE *f) +{ + int r; + stbi__context s; + long pos = ftell(f); + stbi__start_file(&s, f); + r = stbi__is_16_main(&s); + fseek(f,pos,SEEK_SET); + return r; +} +#endif // !STBI_NO_STDIO + +STBIDEF int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp) +{ + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__info_main(&s,x,y,comp); +} + +STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int *x, int *y, int *comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) c, user); + return stbi__info_main(&s,x,y,comp); +} + +STBIDEF int stbi_is_16_bit_from_memory(stbi_uc const *buffer, int len) +{ + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__is_16_main(&s); +} + +STBIDEF int stbi_is_16_bit_from_callbacks(stbi_io_callbacks const *c, void *user) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) c, user); + return stbi__is_16_main(&s); +} + +#endif // STB_IMAGE_IMPLEMENTATION + +/* + revision history: + 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs + 2.19 (2018-02-11) fix warning + 2.18 (2018-01-30) fix warnings + 2.17 (2018-01-29) change sbti__shiftsigned to avoid clang -O2 bug + 1-bit BMP + *_is_16_bit api + avoid warnings + 2.16 (2017-07-23) all functions have 16-bit variants; + STBI_NO_STDIO works again; + compilation fixes; + fix rounding in unpremultiply; + optimize vertical flip; + disable raw_len validation; + documentation fixes + 2.15 (2017-03-18) fix png-1,2,4 bug; now all Imagenet JPGs decode; + warning fixes; disable run-time SSE detection on gcc; + uniform handling of optional "return" values; + thread-safe initialization of zlib tables + 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs + 2.13 (2016-11-29) add 16-bit API, only supported for PNG right now + 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes + 2.11 (2016-04-02) allocate large structures on the stack + remove white matting for transparent PSD + fix reported channel count for PNG & BMP + re-enable SSE2 in non-gcc 64-bit + support RGB-formatted JPEG + read 16-bit PNGs (only as 8-bit) + 2.10 (2016-01-22) avoid warning introduced in 2.09 by STBI_REALLOC_SIZED + 2.09 (2016-01-16) allow comments in PNM files + 16-bit-per-pixel TGA (not bit-per-component) + info() for TGA could break due to .hdr handling + info() for BMP to shares code instead of sloppy parse + can use STBI_REALLOC_SIZED if allocator doesn't support realloc + code cleanup + 2.08 (2015-09-13) fix to 2.07 cleanup, reading RGB PSD as RGBA + 2.07 (2015-09-13) fix compiler warnings + partial animated GIF support + limited 16-bpc PSD support + #ifdef unused functions + bug with < 92 byte PIC,PNM,HDR,TGA + 2.06 (2015-04-19) fix bug where PSD returns wrong '*comp' value + 2.05 (2015-04-19) fix bug in progressive JPEG handling, fix warning + 2.04 (2015-04-15) try to re-enable SIMD on MinGW 64-bit + 2.03 (2015-04-12) extra corruption checking (mmozeiko) + stbi_set_flip_vertically_on_load (nguillemot) + fix NEON support; fix mingw support + 2.02 (2015-01-19) fix incorrect assert, fix warning + 2.01 (2015-01-17) fix various warnings; suppress SIMD on gcc 32-bit without -msse2 + 2.00b (2014-12-25) fix STBI_MALLOC in progressive JPEG + 2.00 (2014-12-25) optimize JPG, including x86 SSE2 & NEON SIMD (ryg) + progressive JPEG (stb) + PGM/PPM support (Ken Miller) + STBI_MALLOC,STBI_REALLOC,STBI_FREE + GIF bugfix -- seemingly never worked + STBI_NO_*, STBI_ONLY_* + 1.48 (2014-12-14) fix incorrectly-named assert() + 1.47 (2014-12-14) 1/2/4-bit PNG support, both direct and paletted (Omar Cornut & stb) + optimize PNG (ryg) + fix bug in interlaced PNG with user-specified channel count (stb) + 1.46 (2014-08-26) + fix broken tRNS chunk (colorkey-style transparency) in non-paletted PNG + 1.45 (2014-08-16) + fix MSVC-ARM internal compiler error by wrapping malloc + 1.44 (2014-08-07) + various warning fixes from Ronny Chevalier + 1.43 (2014-07-15) + fix MSVC-only compiler problem in code changed in 1.42 + 1.42 (2014-07-09) + don't define _CRT_SECURE_NO_WARNINGS (affects user code) + fixes to stbi__cleanup_jpeg path + added STBI_ASSERT to avoid requiring assert.h + 1.41 (2014-06-25) + fix search&replace from 1.36 that messed up comments/error messages + 1.40 (2014-06-22) + fix gcc struct-initialization warning + 1.39 (2014-06-15) + fix to TGA optimization when req_comp != number of components in TGA; + fix to GIF loading because BMP wasn't rewinding (whoops, no GIFs in my test suite) + add support for BMP version 5 (more ignored fields) + 1.38 (2014-06-06) + suppress MSVC warnings on integer casts truncating values + fix accidental rename of 'skip' field of I/O + 1.37 (2014-06-04) + remove duplicate typedef + 1.36 (2014-06-03) + convert to header file single-file library + if de-iphone isn't set, load iphone images color-swapped instead of returning NULL + 1.35 (2014-05-27) + various warnings + fix broken STBI_SIMD path + fix bug where stbi_load_from_file no longer left file pointer in correct place + fix broken non-easy path for 32-bit BMP (possibly never used) + TGA optimization by Arseny Kapoulkine + 1.34 (unknown) + use STBI_NOTUSED in stbi__resample_row_generic(), fix one more leak in tga failure case + 1.33 (2011-07-14) + make stbi_is_hdr work in STBI_NO_HDR (as specified), minor compiler-friendly improvements + 1.32 (2011-07-13) + support for "info" function for all supported filetypes (SpartanJ) + 1.31 (2011-06-20) + a few more leak fixes, bug in PNG handling (SpartanJ) + 1.30 (2011-06-11) + added ability to load files via callbacks to accomidate custom input streams (Ben Wenger) + removed deprecated format-specific test/load functions + removed support for installable file formats (stbi_loader) -- would have been broken for IO callbacks anyway + error cases in bmp and tga give messages and don't leak (Raymond Barbiero, grisha) + fix inefficiency in decoding 32-bit BMP (David Woo) + 1.29 (2010-08-16) + various warning fixes from Aurelien Pocheville + 1.28 (2010-08-01) + fix bug in GIF palette transparency (SpartanJ) + 1.27 (2010-08-01) + cast-to-stbi_uc to fix warnings + 1.26 (2010-07-24) + fix bug in file buffering for PNG reported by SpartanJ + 1.25 (2010-07-17) + refix trans_data warning (Won Chun) + 1.24 (2010-07-12) + perf improvements reading from files on platforms with lock-heavy fgetc() + minor perf improvements for jpeg + deprecated type-specific functions so we'll get feedback if they're needed + attempt to fix trans_data warning (Won Chun) + 1.23 fixed bug in iPhone support + 1.22 (2010-07-10) + removed image *writing* support + stbi_info support from Jetro Lauha + GIF support from Jean-Marc Lienher + iPhone PNG-extensions from James Brown + warning-fixes from Nicolas Schulz and Janez Zemva (i.stbi__err. Janez (U+017D)emva) + 1.21 fix use of 'stbi_uc' in header (reported by jon blow) + 1.20 added support for Softimage PIC, by Tom Seddon + 1.19 bug in interlaced PNG corruption check (found by ryg) + 1.18 (2008-08-02) + fix a threading bug (local mutable static) + 1.17 support interlaced PNG + 1.16 major bugfix - stbi__convert_format converted one too many pixels + 1.15 initialize some fields for thread safety + 1.14 fix threadsafe conversion bug + header-file-only version (#define STBI_HEADER_FILE_ONLY before including) + 1.13 threadsafe + 1.12 const qualifiers in the API + 1.11 Support installable IDCT, colorspace conversion routines + 1.10 Fixes for 64-bit (don't use "unsigned long") + optimized upsampling by Fabian "ryg" Giesen + 1.09 Fix format-conversion for PSD code (bad global variables!) + 1.08 Thatcher Ulrich's PSD code integrated by Nicolas Schulz + 1.07 attempt to fix C++ warning/errors again + 1.06 attempt to fix C++ warning/errors again + 1.05 fix TGA loading to return correct *comp and use good luminance calc + 1.04 default float alpha is 1, not 255; use 'void *' for stbi_image_free + 1.03 bugfixes to STBI_NO_STDIO, STBI_NO_HDR + 1.02 support for (subset of) HDR files, float interface for preferred access to them + 1.01 fix bug: possible bug in handling right-side up bmps... not sure + fix bug: the stbi__bmp_load() and stbi__tga_load() functions didn't work at all + 1.00 interface to zlib that skips zlib header + 0.99 correct handling of alpha in palette + 0.98 TGA loader by lonesock; dynamically add loaders (untested) + 0.97 jpeg errors on too large a file; also catch another malloc failure + 0.96 fix detection of invalid v value - particleman@mollyrocket forum + 0.95 during header scan, seek to markers in case of padding + 0.94 STBI_NO_STDIO to disable stdio usage; rename all #defines the same + 0.93 handle jpegtran output; verbose errors + 0.92 read 4,8,16,24,32-bit BMP files of several formats + 0.91 output 24-bit Windows 3.0 BMP files + 0.90 fix a few more warnings; bump version number to approach 1.0 + 0.61 bugfixes due to Marc LeBlanc, Christopher Lloyd + 0.60 fix compiling as c++ + 0.59 fix warnings: merge Dave Moore's -Wall fixes + 0.58 fix bug: zlib uncompressed mode len/nlen was wrong endian + 0.57 fix bug: jpg last huffman symbol before marker was >9 bits but less than 16 available + 0.56 fix bug: zlib uncompressed mode len vs. nlen + 0.55 fix bug: restart_interval not initialized to 0 + 0.54 allow NULL for 'int *comp' + 0.53 fix bug in png 3->4; speedup png decoding + 0.52 png handles req_comp=3,4 directly; minor cleanup; jpeg comments + 0.51 obey req_comp requests, 1-component jpegs return as 1-component, + on 'test' only check type, not whether we support this variant + 0.50 (2006-11-19) + first released version +*/ + + +/* +------------------------------------------------------------------------------ +This software is available under 2 licenses -- choose whichever you prefer. +------------------------------------------------------------------------------ +ALTERNATIVE A - MIT License +Copyright (c) 2017 Sean Barrett +Permission is hereby granted, free of charge, to any person obtaining a copy of +this software and associated documentation files (the "Software"), to deal in +the Software without restriction, including without limitation the rights to +use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies +of the Software, and to permit persons to whom the Software is furnished to do +so, subject to the following conditions: +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. +------------------------------------------------------------------------------ +ALTERNATIVE B - Public Domain (www.unlicense.org) +This is free and unencumbered software released into the public domain. +Anyone is free to copy, modify, publish, use, compile, sell, or distribute this +software, either in source code form or as a compiled binary, for any purpose, +commercial or non-commercial, and by any means. +In jurisdictions that recognize copyright laws, the author or authors of this +software dedicate any and all copyright interest in the software to the public +domain. We make this dedication for the benefit of the public at large and to +the detriment of our heirs and successors. We intend this dedication to be an +overt act of relinquishment in perpetuity of all present and future rights to +this software under copyright law. +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN +ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +------------------------------------------------------------------------------ +*/ \ No newline at end of file diff --git a/src/lib/thread.c b/src/lib/thread.c new file mode 100644 index 0000000..e081288 --- /dev/null +++ b/src/lib/thread.c @@ -0,0 +1,109 @@ +#include "thread.h" + +typedef struct ThreadQueue { + void* items; + u32 type_size; + u32 max; + u32 head; + u32 tail; + u32 count; + Mutex mutex; + Cond not_empty; + Cond not_full; +} ThreadQueue; + +Thread spawnThread(void * (*threadFn)(void *), void* thread_arg) { + pthread_t thread; + pthread_create(&thread, NULL, threadFn, thread_arg); + Thread result = { thread }; + return result; +} + +Mutex newMutex() { + Mutex result = { PTHREAD_MUTEX_INITIALIZER }; + return result; +} + +Cond newCond() { + Cond result = { 0 }; + pthread_cond_init(&result.cond, NULL); + return result; +} + +void lockMutex(Mutex* m) { + pthread_mutex_lock(&m->mutex); +} + +void unlockMutex(Mutex* m) { + pthread_mutex_unlock(&m->mutex); +} + +void signalCond(Cond* cond) { + pthread_cond_signal(&cond->cond); +} + +void waitForCondSignal(Cond* cond, Mutex* mutex) { + pthread_cond_wait(&cond->cond, &mutex->mutex); +} + +fn ThreadQueue newThreadQueue(Arena* a, u32 type_size, u32 items_max) { + ThreadQueue result = {0}; + result.type_size = type_size; + result.max = items_max; + result.items = arenaAllocArraySized(a, type_size, items_max); + result.mutex = newMutex(); + result.not_full = newCond(); + result.not_empty = newCond(); + return result; +} + +fn void threadSafeQueuePush(ThreadQueue* queue, void* item) { + lockMutex(&queue->mutex); { + while (queue->count == queue->max) { + waitForCondSignal(&queue->not_full, &queue->mutex); + } + + memcpy(queue->items + (queue->tail * (sizeof item)), item, queue->type_size); + queue->tail = (queue->tail + 1) % queue->max; + queue->count++; + + signalCond(&queue->not_empty); + } unlockMutex(&queue->mutex); +} + +fn void* threadSafeQueuePop(ThreadQueue* q) { + void* result = NULL; + + lockMutex(&q->mutex); { + while (q->count == 0) { + waitForCondSignal(&q->not_empty, &q->mutex); + } + + result = &q->items[q->head]; + q->head = (q->head + 1) % q->max; + q->count--; + + signalCond(&q->not_full); + } unlockMutex(&q->mutex); + + return result; +} + +// immediately returns NULL if there's nothing in the ThreadQueue +fn void* threadSafeNonblockingQueuePop(ThreadQueue* q, void* copy_target, u64 len) { + void* result = NULL; + + lockMutex(&q->mutex); { + if (q->count > 0) { + result = &q->items[q->head]; + MemoryCopy(copy_target, result, len); + q->head = (q->head + 1) % q->max; + q->count--; + + signalCond(&q->not_full); + } + } unlockMutex(&q->mutex); + + return result; +} + diff --git a/src/lib/thread.h b/src/lib/thread.h new file mode 100644 index 0000000..c920975 --- /dev/null +++ b/src/lib/thread.h @@ -0,0 +1,29 @@ +#ifndef LIB_THREAD_H +#define LIB_THREAD_H + +#include +#include +#include +#include "../base/include.h" + +typedef struct Thread { + pthread_t thread; +} Thread; + +typedef struct Mutex { + pthread_mutex_t mutex; +} Mutex; + +typedef struct Cond { + pthread_cond_t cond; +} Cond; + +Thread spawnThread(void * (*threadFn)(void *), void* thread_arg); +Mutex newMutex(); +Cond newCond(); +void lockMutex(Mutex* m); +void unlockMutex(Mutex* m); +void signalCond(Cond* cond); +void waitForCondSignal(Cond* cond, Mutex* mutex); + +#endif //LIB_THREAD_H diff --git a/src/lib/tui.c b/src/lib/tui.c new file mode 100644 index 0000000..2b10d8c --- /dev/null +++ b/src/lib/tui.c @@ -0,0 +1,579 @@ +#include "../base/all.h" +#include "../string_chunk.h" +#include +#include +#include +#include + +#define UTF8_MAX_WIDTH 4 +#define ANSI_HP_RED (196) +#define ANSI_MP_BLUE (33) +#define ANSI_LIGHT_GREEN (82) +#define ANSI_MID_GREEN (70) +#define ANSI_DARK_GREEN (35) +#define ANSI_BLACK (16) +#define ANSI_WHITE (15) +#define ANSI_BROWN (130) +#define ANSI_DULL_RED (1) +#define ANSI_HIGHLIGHT_RED (9) +#define ANSI_DULL_GREEN (2) +#define ANSI_HIGHLIGHT_GREEN (10) +#define ANSI_DULL_YELLOW (3) +#define ANSI_HIGHLIGHT_YELLOW (11) +#define ANSI_DULL_BLUE (4) +#define ANSI_HIGHLIGHT_BLUE (12) +#define ANSI_DULL_GRAY (7) +#define ANSI_HIGHLIGHT_GRAY (16) +#define MAX_COMMAND_PALETTE_COMMANDS (1000) + +///// TYPES +typedef struct Pixel { + u8 foreground; + u8 background; + u8 bytes[UTF8_MAX_WIDTH]; +} Pixel; + +typedef struct TuiState { + bool redraw; + ptr writeable_output_ansi_string; + Pixel* frame_buffer; + Pixel* back_buffer; + u64 buffer_len; // how many Pixels + Pos2 cursor; + Pos2 prev_cursor; + Dim2 screen_dimensions; + Dim2 prev_screen_dimensions; +} TuiState; + +typedef struct RGB { + u8 r; + u8 g; + u8 b; +} RGB; + +typedef struct CommandPaletteCommand { + u32 id; + ptr display_name; + ptr description; + ptr* tags; +} CommandPaletteCommand; + +typedef struct CommandPaletteCommandList { + u32 length; + CommandPaletteCommand* items; +} CommandPaletteCommandList; + +typedef struct StringSearchScore { + bool name_matched; + bool description_matched; + u16 tag_match_count; + u32 score; + u32 name_match_start; + u32 name_match_len; + u32 description_match_start; + u32 description_match_len; +} StringSearchScore; + +///// Functions() +fn u32 rgbToNum(RGB rgb) { + return ((rgb.r<<16) | (rgb.g<<8) | rgb.b); +} + +fn u8 rgbToAnsi(RGB color) { + u8 r = color.r / (255 / 5); + u8 g = color.g / (255 / 5); + u8 b = color.b / (255 / 5); + return 16 + (36*r) + (6*g) + b; +} + +fn RGB rgbDarken(RGB color, f32 factor) { + RGB result = { + .r = ((f32)color.r) * factor, + .g = ((f32)color.g) * factor, + .b = ((f32)color.b) * factor, + }; + return result; +} + +/* +fn RGB ansiToRGB(u8 ansi) { + ansi -= 16; + u8 b = ansi % 51; + ansi -= () + u8 g = +} +*/ + +fn bool rgbEq(RGB a, RGB b) { + return a.r == b.r && a.g == b.g && a.b == b.b; +} + +fn bool isPixelEq(Pixel a, Pixel b) { + return a.background == b.background + && a.foreground == b.foreground + && a.bytes[0] == b.bytes[0] + && a.bytes[1] == b.bytes[1] + && a.bytes[2] == b.bytes[2] + && a.bytes[3] == b.bytes[3] + ; +} + +fn TuiState tuiInit(Arena* a, u64 buffer_len) { + TuiState result = { + .redraw = false, + .writeable_output_ansi_string = arenaAlloc(a, MB(1)), + .buffer_len = buffer_len, + .back_buffer = arenaAllocArray(a, Pixel, buffer_len), // allocate biggest possible dimensions + .frame_buffer = arenaAllocArray(a, Pixel, buffer_len), // allocate biggest possible dimensions + }; + MemoryZero(result.back_buffer, buffer_len * sizeof(Pixel)); + MemoryZero(result.frame_buffer, buffer_len * sizeof(Pixel)); + return result; +} + +fn void copyStr(u8* bytes, str cstring) { + for (u32 i = 0; i < strlen(cstring); i++) { + bytes[i] = cstring[i]; + } +} + +fn void drawAnsiBox(Pixel* buf, Box box, Dim2 sd, bool bold) { + str items[] = {"┌","─","┐","│","└","┘"}; + str b_items[] = {"┏","━","┓","┃","┗","┛"}; + ptr* use = (ptr*)items; + if (bold) { + use = (ptr*)b_items; + } + u32 pos = XYToPos(box.x, box.y, sd.width); + + // print the upper box border + copyStr(buf[pos].bytes, use[0]); + for (i32 i = 0; i < box.width; i++) { + copyStr(buf[pos+1+i].bytes, use[1]); + } + copyStr(buf[pos+1+box.width].bytes, use[2]); + + // start printing the rows + for (i32 i = 0; i < box.height; i++) { + pos = XYToPos(box.x, (box.y+1+i), sd.width); // move cursor to beginning of the row + copyStr(buf[pos].bytes, use[3]); + copyStr(buf[pos+1+box.width].bytes, use[3]); + } + + // print the bottom box border + pos = XYToPos(box.x, (box.y+1+box.height), sd.width); + copyStr(buf[pos].bytes, use[4]); + for (i32 i = 0; i < box.width; i++) { + copyStr(buf[pos+1+i].bytes, use[1]); + } + copyStr(buf[pos+1+box.width].bytes, use[5]); +} + +fn void renderUtf8CharToBuffer(Pixel* buf, u16 x, u16 y, str text, Dim2 screen_dimensions) { + assert(strlen(text) <= UTF8_MAX_WIDTH); + u32 pos = x + (screen_dimensions.width*y); + for (u32 i = 0; i < strlen(text); i++) { + buf[pos].bytes[i] = text[i]; + } +} + +fn void renderStrToBufferMaxWidth(Pixel* buf, u16 x, u16 y, str text, u16 width, Dim2 screen_dimensions) { + u32 pos = x + (screen_dimensions.width*y); + for (u32 i = 0; i < strlen(text); i++) { + buf[pos + (i % width)].background = 0; + buf[pos + (i % width)].foreground = 0; + buf[pos + (i % width)].bytes[0] = text[i]; + if (i % width == (width-1)) { // on last char i inside width + pos += screen_dimensions.width; // move pos to next line + } + } +} + +fn void renderStrToBufferMaxWidthWithoutChangingColor(Pixel* buf, u16 x, u16 y, str text, u16 width, Dim2 screen_dimensions) { + u32 pos = x + (screen_dimensions.width*y); + for (u32 i = 0; i < strlen(text); i++) { + buf[pos + (i % width)].bytes[0] = text[i]; + if (i % width == (width-1)) { // on last char i inside width + pos += screen_dimensions.width; // move pos to next line + } + } +} + +fn void renderStrToBuffer(Pixel* buf, u16 x, u16 y, str text, Dim2 screen_dimensions) { + u32 pos = x + (screen_dimensions.width*y); + for (u32 i = 0; i < strlen(text); i++) { + buf[pos+i].bytes[0] = text[i]; + } +} + +fn void renderStringChunkList(TuiState* tui, StringChunkList* list, u16 x, u16 y) { + u32 pos = XYToPos(x, y, tui->screen_dimensions.width); + StringChunk* chunk = list->first; + for (u32 i = 0; i < list->total_size; i++) { + if (i > 0 && i % STRING_CHUNK_PAYLOAD_SIZE == 0) { + chunk = chunk->next; + } + tui->frame_buffer[pos+i].bytes[0] = *((char*)(chunk + 1) + (i%STRING_CHUNK_PAYLOAD_SIZE)); + } +} + +fn u32 sprintfAnsiMoveCursorTo(ptr output, u16 x, u16 y) { + return sprintf(output, "\x1b[%d;%df",y,x); +} + +fn void printfBufferAndSwap(TuiState* tui) { + Pixel* old = tui->back_buffer; + Pixel* next = tui->frame_buffer; + u32 length = tui->screen_dimensions.height * tui->screen_dimensions.width; + bool screen_dimensions_changed = tui->screen_dimensions.height != tui->prev_screen_dimensions.height + || tui->screen_dimensions.width != tui->prev_screen_dimensions.width; + bool should_redraw_whole_screen = screen_dimensions_changed || tui->redraw; + + // "quick" exit this fn if old == next + if (!should_redraw_whole_screen) { + bool old_equals_new = true; + for (u32 i = 0; i < length; i++) { + if (old[i].background != next[i].background + || old[i].foreground != next[i].foreground + || old[i].bytes[0] != next[i].bytes[0] + || old[i].bytes[1] != next[i].bytes[1] + || old[i].bytes[2] != next[i].bytes[2] + || old[i].bytes[3] != next[i].bytes[3]) { + old_equals_new = false; + break; + } + } + if (old_equals_new + && tui->prev_cursor.x == tui->cursor.x + && tui->prev_cursor.y == tui->cursor.y + ) { + length += 0; // debugging helper line + return; // skip all the write() and sprintf() calls, since the frames are the same. + } + } + + ptr output = tui->writeable_output_ansi_string; + u8 bg = 0; + u8 fg = 0; + u16 x = 1; + u16 y = 1; + u16 last_x = 0; + u16 last_y = 0; + str RESET_STYLES = "\033[0m"; + str CLEAR_SCREEN = "\033[2J"; + + if (should_redraw_whole_screen) { + output += sprintf(output, "\033[0m\033[2J"); + output += sprintfAnsiMoveCursorTo(output, x, y); + bool printed_last = false; + for (u32 i = 0; i < length; i++) { + x = (i % tui->screen_dimensions.width) + 1; + y = (i / tui->screen_dimensions.width) + 1; + if (next[i].bytes[0] != 0) { + if (printed_last == false) { + output += sprintfAnsiMoveCursorTo(output, x, y); + } + char bytes[5] = {0}; // do this nonsense to ensure null-terminated characters + bytes[0] = next[i].bytes[0]; + bytes[1] = next[i].bytes[1]; + bytes[2] = next[i].bytes[2]; + bytes[3] = next[i].bytes[3]; + + if (next[i].background != bg || next[i].foreground != fg) { + bg = next[i].background; + fg = next[i].foreground; + if (bg == 0 && fg == 0) { + output += sprintf(output, "%s%s", RESET_STYLES, bytes); + } else if (bg != 0 && fg == 0) { + output += sprintf(output, "\033[48;5;%dm%s", bg, bytes); + } else if (bg == 0 && fg != 0) { + output += sprintf(output, "\033[38;5;%dm%s", fg, bytes); + } else { + output += sprintf(output, "\033[48;5;%d;38;5;%dm%s", bg, fg, bytes); + } + } else { + output += sprintf(output, "%s", bytes); + } + printed_last = true; + } else { + printed_last = false; + } + } + } else { + // clearing pass, to overwrite things that were there on the last frame, but are no longer present + // we do this before the "rendering" pass so that multi-space characters (emojis) are easier to deal with + output += sprintf(output, "%s", RESET_STYLES); + output += sprintfAnsiMoveCursorTo(output, x, y); + for (u32 i = 0; i < length; i++) { + x = (i % tui->screen_dimensions.width) + 1; + y = (i / tui->screen_dimensions.width) + 1; + bool needs_clearing = (next[i].bytes[0] == 0 && old[i].bytes[0] != 0) + || (next[i].background == 0 && old[i].background != 0) + || (next[i].foreground == 0 && old[i].foreground != 0); + if (needs_clearing) { + bool last_printed_pos_is_adjacent_to_current_x_y = last_x+1 == x && last_y == y; + if (!last_printed_pos_is_adjacent_to_current_x_y) { + output += sprintfAnsiMoveCursorTo(output, x, y); + } + output += sprintf(output, " "); + last_x = x; + last_y = y; + } + } + + // rendering pass + last_x = 0; + last_y = 0; + for (u32 i = 0; i < length; i++) { + x = (i % tui->screen_dimensions.width) + 1; + y = (i / tui->screen_dimensions.width) + 1; + if (!isPixelEq(old[i], next[i])) { + if (next[i].bytes[0] != 0) { + bool last_printed_pos_is_adjacent_to_current_x_y = last_x+1 == x && last_y == y; + if (!last_printed_pos_is_adjacent_to_current_x_y) { + output += sprintfAnsiMoveCursorTo(output, x, y); + } + + u8 bytes[5] = {0}; // do this nonsense to ensure null-terminated characters + bytes[0] = next[i].bytes[0]; + bytes[1] = next[i].bytes[1]; + bytes[2] = next[i].bytes[2]; + bytes[3] = next[i].bytes[3]; + + if (next[i].background != bg || next[i].foreground != fg) { + bg = next[i].background; + fg = next[i].foreground; + if (bg == 0 && fg == 0) { + output += sprintf(output, "%s%s", RESET_STYLES, bytes); + } else if (bg != 0 && fg == 0) { + output += sprintf(output, "\033[48;5;%dm%s", bg, bytes); + } else if (bg == 0 && fg != 0) { + output += sprintf(output, "\033[38;5;%dm%s", fg, bytes); + } else { + output += sprintf(output, "\033[48;5;%d;38;5;%dm%s", bg, fg, bytes); + } + } else { + output += sprintf(output, "%s", bytes); + } + + last_x = x; + last_y = y; + } + } + } + } + + output += sprintfAnsiMoveCursorTo(output, tui->cursor.x+1, tui->cursor.y+1); // re-position the cursor according to what the rendering logic set it to + + // finally write our whole string to the terminal + i64 count = output - tui->writeable_output_ansi_string; + osBlitToTerminal(tui->writeable_output_ansi_string, count); + + // swap our buffers + Pixel* tmp = tui->back_buffer; + tui->back_buffer = tui->frame_buffer; + tui->frame_buffer = tmp; + + // reset the `redraw` flag + tui->redraw = false; +} + +fn u32 matchCommandPaletteCommands(String current_search, CommandPaletteCommandList commands, u32 menu_index, u32* scores, StringSearchScore* score_details) { + // returns the `commands` id that matches the `menu_index` + for (u32 i = 0; i < commands.length; i++) { + CommandPaletteCommand* cmd = &commands.items[i]; + bool name_matches = false; + for (u32 j = 0; j < strlen(cmd->display_name); j++) { + if (current_search.length > 0 && lowerAscii(cmd->display_name[j]) == lowerAscii(current_search.bytes[0])) { + for (u32 k = 0; k < current_search.length && k+j < strlen(cmd->display_name); k++) { + if (lowerAscii(current_search.bytes[k]) != lowerAscii(cmd->display_name[j+k])) { + break; + } + if (k == current_search.length - 1) { + name_matches = true; + score_details[i].name_match_start = j; + score_details[i].name_match_len = current_search.length; + } + } + } + } + bool description_matches = false; + u32 tag_match_count = 0; + scores[i] = (name_matches * 100 * MAX_COMMAND_PALETTE_COMMANDS) + + (description_matches * 10 * MAX_COMMAND_PALETTE_COMMANDS) + + (tag_match_count * MAX_COMMAND_PALETTE_COMMANDS) + + cmd->id; + score_details[i].score = scores[i]; + score_details[i].name_matched = name_matches; + score_details[i].description_matched = description_matches; + score_details[i].tag_match_count = tag_match_count; + } + u32Quicksort(scores, 0, commands.length - 1); + u32ReverseArray(scores, commands.length); + return (scores[menu_index] % MAX_COMMAND_PALETTE_COMMANDS); +} + +fn Pos2 renderCommandPalette(TuiState* tui, String current_search, CommandPaletteCommandList commands, u32 menu_index) { + // returns the cursor position as Pos2 + + ScratchMem scratch = scratchGet(); + Dim2 sd = tui->screen_dimensions; + Pos2 result = { 0 }; + + // draw the outline + Box outline = { + .width = sd.width / 2, + .height = sd.height / 2, + }; + outline.x = outline.width / 2; + outline.y = outline.height / 2; + drawAnsiBox(tui->frame_buffer, outline, sd, true); + + // draw the "search bar" + result.x = outline.x + 1 + current_search.length; + result.y = outline.y + 1; + renderStrToBufferMaxWidth(tui->frame_buffer, outline.x+1, outline.y+1, current_search.bytes, outline.width - 2, sd); + for (u32 i = 1; i < outline.width-1; i++) { + u32 pos = XYToPos(outline.x+1, outline.y+2, sd.width); + copyStr(tui->frame_buffer[pos].bytes, "━"); + } + + // sort the command options + u32* scores = arenaAllocArray(&scratch.arena, u32, commands.length); + StringSearchScore* score_details = arenaAllocArray(&scratch.arena, StringSearchScore, commands.length); + matchCommandPaletteCommands(current_search, commands, menu_index, scores, score_details); + + // draw the command options + u32 x = outline.x + 1; + u32 y = outline.y + 3; + for (u32 i = 0; (y-outline.y) < (outline.height-1) && i < commands.length; i++, y+=2) { + if (i == menu_index) { + for (u32 ii = 0; ii < outline.width-1; ii++) { + u32 pos = XYToPos(x+ii, y, tui->screen_dimensions.width); + tui->frame_buffer[pos].bytes[0] = ' '; + tui->frame_buffer[pos].foreground = ANSI_BLACK; + tui->frame_buffer[pos].background = ANSI_WHITE; + pos = XYToPos(x+ii, y+1, tui->screen_dimensions.width); + tui->frame_buffer[pos].bytes[0] = ' '; + tui->frame_buffer[pos].foreground = ANSI_BLACK; + tui->frame_buffer[pos].background = ANSI_WHITE; + } + } + u32 id = scores[i] % MAX_COMMAND_PALETTE_COMMANDS; + StringSearchScore score = score_details[id]; + CommandPaletteCommand* cmd = NULL; + for (u32 j = 0; j < commands.length; j++) { + if (commands.items[j].id == id) { + cmd = &commands.items[j]; + break; + } + } + assert(cmd != NULL); + if (score.name_matched) { + u32 end_idx = score.name_match_len + score.name_match_start; + for (u32 ii = 0; ii < strlen(cmd->display_name); ii++) { + u32 pos = XYToPos(x+ii, y, tui->screen_dimensions.width); + if (ii >= score.name_match_start && ii < end_idx) { + if (i == menu_index) { + tui->frame_buffer[pos].foreground = ANSI_DULL_RED; + } else { + tui->frame_buffer[pos].foreground = ANSI_HIGHLIGHT_YELLOW; + tui->frame_buffer[pos].background = 0; + } + } + tui->frame_buffer[pos].bytes[0] = cmd->display_name[ii]; + } + } else { + renderStrToBufferMaxWidthWithoutChangingColor(tui->frame_buffer, x, y, cmd->display_name, outline.width - 2, sd); + } + renderStrToBufferMaxWidthWithoutChangingColor(tui->frame_buffer, x, y+1, cmd->description, outline.width - 2, sd); + } + + scratchReturn(&scratch); + return result; +} + +fn void renderChoiceMenu(TuiState* tui, u16 x, u16 y, ptr options[], u32 len, bool choosable, u32 selected_index, u8* colors) { + for (u32 i = 0; i < len; i++) { + u32 pos = x + (tui->screen_dimensions.width*(y+i)); + if (choosable && selected_index == i) { + tui->cursor.x = x; + tui->cursor.y = y+i; + } + u8 bytes[80] = {0}; + sprintf((char*)bytes, "- %d. ", i+1); + u32 offset = strlen((char*)bytes); + u32 bytes_remaining = (80 - offset); + for ( + u32 j = 0; + j < strlen(options[i]) && j < bytes_remaining; + j++ + ) { + bytes[offset+j] = options[i][j]; + } + // write our `bytes` buffer into the Pixel* buf + for (u32 j = 0; j < 80; j++) { + tui->frame_buffer[pos+j].bytes[0] = bytes[j]; + if (choosable && selected_index == i) { + tui->frame_buffer[pos+j].foreground = ANSI_BLACK; + tui->frame_buffer[pos+j].background = ANSI_WHITE; + } else { + tui->frame_buffer[pos+j].foreground = colors == NULL ? ANSI_WHITE : colors[i]; + tui->frame_buffer[pos+j].background = ANSI_BLACK; + } + } + } +} + +fn void infiniteUILoop( + u32 max_screen_width, + u32 max_screen_height, + u64 goal_input_loop_us, + void* state, + // updateAndRender should return a bool `should_quit` + bool (*updateAndRender)(TuiState* tui, void* state, u8* input_buffer, u64 loop_count) +) { + bool should_quit = false; + Arena permanent_arena = {0}; + arenaInit(&permanent_arena); + // set up the TUI incantations + TermIOs old_terminal_attributes = osStartTUI(false); + TuiState tui = tuiInit(&permanent_arena, max_screen_width*max_screen_height); + tui.screen_dimensions = osGetTerminalDimensions(); + + // ui loop (read input, simulate next frame, render) + u8 input_buffer[5] = {0}; + u64 loop_start; + u64 loop_count = 0; + while (!should_quit) { + loop_count += 1; + loop_start = osTimeMicrosecondsNow(); + + // both reads local input from the keyboard AND from the network + osReadConsoleInput(input_buffer, 4); + + // prep rendering + MemoryZero(tui.frame_buffer, tui.buffer_len * sizeof(Pixel)); + tui.prev_screen_dimensions = tui.screen_dimensions; + if (loop_count % 17 == 0) { // every 17 frames, update our terminal_dimensions + tui.screen_dimensions = osGetTerminalDimensions(); + } + tui.prev_cursor = tui.cursor;// save last frame's cursor + + // operate on input + render new tui.frame_buffer + should_quit = updateAndRender(&tui, state, input_buffer, loop_count); + + printfBufferAndSwap(&tui); + + // loop timing + u32 loop_duration = osTimeMicrosecondsNow() - loop_start; + i32 remaining_time = goal_input_loop_us - loop_duration; + if (remaining_time > 0) { + osSleepMicroseconds(remaining_time); + } + } + + // cleanup terminal TUI incantations + osEndTUI(old_terminal_attributes); +} + diff --git a/src/render.c b/src/render.c new file mode 100644 index 0000000..4a1d009 --- /dev/null +++ b/src/render.c @@ -0,0 +1,104 @@ +#include "shared.h" +#include "lib/tui.c" + +#define MAX_SCREEN_HEIGHT 300 +#define MAX_SCREEN_WIDTH 800 + +fn str charForEntity(EntityType e) { + switch (e) { + case EntityWall: + return "# "; + case EntityDoor: + return "🚪"; + case EntityCharacter: + return "🧙"; + //return "웃"; + case EntityNull: + case EntityType_Count: + //default: // commented out so that we get a warning for missing entity + return " "; + } +} + +/* HERE IS WHERE I USUALLY PUT THE "room" OR "map" DRAWING FUNCTION + * +fn void renderRoom(Pixel* buf, u32 x, u32 y, RenderableRoom* room, Dim2 screen_dimensions, bool active) { + // x,y is starting cursor location of upper-left corner + + Box b = { .x = x, .y = y, .width = ROOM_WIDTH*2, .height = ROOM_HEIGHT }; + drawAnsiBox(buf, b, screen_dimensions, active); + + // start printing the rows + // move cursor to beginning of the room + for (i32 j = 0; j < ROOM_HEIGHT; j++) { + for (i32 i = 0; i < ROOM_WIDTH; i++) { + u32 roompos = XYToPos(i, j, ROOM_WIDTH); + if (!room->visible[roompos] && room->memory[roompos] == RememberedTileQualityNone) { + continue; + } + u32 bufpos = (x+1+(i*2)) + (screen_dimensions.width*(y+1+j)); + + str fg_char = charForEntity(room->foreground[roompos]); + RGB background_color = colorForTile(room->background[roompos]); + if (room->visible[roompos]) { + buf[bufpos].foreground = ansiColorForEntity(room->foreground[roompos]); + if (room->light[roompos] < VISIBLE_BRIGHTNESS_CUTOFF) { + fg_char = charForEntity(EntityMurkyUnknown); + background_color = rgbDarken(background_color, 0.8); + } + } else if (room->memory[roompos] == RememberedTileQualityDim) { + background_color = rgbDarken(background_color, 0.4); + buf[bufpos].foreground = ansiColorForEntity(room->foreground[roompos]); + fg_char = charForEntity(EntityMurkyUnknown); + } else if (room->memory[roompos] == RememberedTileQualityClear) { + background_color = rgbDarken(background_color, 0.6); + buf[bufpos].foreground = ansiColorForEntity(room->foreground[roompos]); + } + buf[bufpos].background = rgbToAnsi(background_color); + Utf8Character first_character_class = classifyUtf8Character((u8)fg_char[0]); + // if it's a single ASCII character + if (first_character_class == Utf8CharacterAscii && fg_char[1] == '\0') { + buf[bufpos].bytes[0] = fg_char[0]; + // if it's a pair of ASCII characters + } else if (first_character_class == Utf8CharacterAscii && fg_char[1] > 0 && fg_char[2] == '\0') { + buf[bufpos].bytes[0] = fg_char[0]; + buf[bufpos+1].bytes[0] = fg_char[1]; + buf[bufpos+1].background = buf[bufpos].background; + buf[bufpos+1].foreground = buf[bufpos].foreground; + } else if (first_character_class == Utf8CharacterThreeByte) { + if (strlen(fg_char) == 3) {// handle 2-wide characters + buf[bufpos+1].background = buf[bufpos].background; + buf[bufpos+1].foreground = buf[bufpos].foreground; + } + for (u32 k = 0; k < strlen(fg_char)/3; k++) { + if (k != 0) { + buf[bufpos+k].background = buf[bufpos].background; + buf[bufpos+k].foreground = buf[bufpos].foreground; + } + for (u32 l = 0; l < 3; l++) { + buf[bufpos+k].bytes[l] = fg_char[l+(k*3)]; + } + } + } else if (first_character_class == Utf8CharacterFourByte) { + if (strlen(fg_char) == UTF8_MAX_WIDTH) {// assume all of these characters are 2-wide + buf[bufpos+1].background = buf[bufpos].background; + buf[bufpos+1].foreground = buf[bufpos].foreground; + } + for (u32 k = 0; k < strlen(fg_char)/UTF8_MAX_WIDTH; k++) { + for (u32 l = 0; l < UTF8_MAX_WIDTH; l++) { + buf[bufpos+k].bytes[l] = fg_char[l+(k*UTF8_MAX_WIDTH)]; + } + } + // handle secondary bytes that didn't divide evenly + for (u32 k = 0; k < strlen(fg_char)%UTF8_MAX_WIDTH; k++) { + buf[bufpos+1].background = buf[bufpos].background; + buf[bufpos+1].foreground = buf[bufpos].foreground; + buf[bufpos+1].bytes[k] = fg_char[UTF8_MAX_WIDTH+k]; + } + } else { + assert(false && "unhandled bullshit"); + } + } + } +} +*/ diff --git a/src/server.c b/src/server.c new file mode 100644 index 0000000..3bdef07 --- /dev/null +++ b/src/server.c @@ -0,0 +1,785 @@ +/* + * Code conventions: + * MyStructType + * myFunction() + * MyMacro() + * my_variable + * MY_CONSTANT + * */ +#include +#include +#include +#include "shared.h" +#include "base/impl.c" +#define NET_OUTGOING_MESSAGE_QUEUE_LEN 64 +#include "lib/network.c" +#include "render.c" +#include "string_chunk.c" + +///// CONSTANTS +#define MAX_ENTITIES (2<<18) +#define LEFT_ROOM_ENTITES_LEN (KB(1)) +#define ROOM_MAP_COLLISIONS_LEN MAX_ROOMS/8 +#define CLIENT_COMMAND_LIST_LEN 8 +#define SERVER_PORT 7777 +#define SERVER_MAX_HEAP_MEMORY MB(256) +#define SERVER_MAX_CLIENTS 16 +#define GAME_THREAD_CONCURRENCY 4 +#define GOAL_NETWORK_SEND_LOOPS_PER_S 4 +#define GOAL_NETWORK_SEND_LOOP_US 1000000/GOAL_NETWORK_SEND_LOOPS_PER_S +#define GOAL_GAME_LOOPS_PER_S 4 +#define GOAL_GAME_LOOP_US 1000000/GOAL_GAME_LOOPS_PER_S +#define CLIENT_TIMEOUT_FRAMES GOAL_GAME_LOOPS_PER_S*3 +#define CHUNK_SIZE 64 +#define ACCOUNT_CHUNK_SIZE 64 +#define PARSED_CLIENT_COMMAND_THREAD_QUEUE_LEN 64 + +///// TypeDefs +typedef struct ParsedClientCommand { + CommandType type; + u8 byte; + u16 sender_port; + u16 alt_port; + u32 sender_ip; + u32 alt_ip; + StringChunkList name; + StringChunkList pass; + u64 id; +} ParsedClientCommand; + +typedef struct ParsedClientCommandThreadQueue { + ParsedClientCommand items[PARSED_CLIENT_COMMAND_THREAD_QUEUE_LEN]; + u32 head; + u32 tail; + u32 count; + Mutex mutex; + Cond not_empty; + Cond not_full; +} ParsedClientCommandThreadQueue; + +typedef struct Entity { + bool changed; + u8 x; + u8 y; + u8 color; + EntityType type; + u32 misc; + u64 id; + u64 features; +} Entity; + +typedef struct Account { + u64 id; + String name; + String pw; + u64 eid; +} Account; + +typedef struct AccountChunk { + u64 length; // the currently used # of accounts in this chunk + u64 capacity; // the "chunk size" / space in this chunk + struct AccountChunk* next; // the next chunk + Account* items; // the actual accounts +} AccountChunk; + +typedef struct EntityList { + u64 length; // the currently used length + u64 capacity; + Entity* items; +} EntityList; + +typedef struct EntityChunk { + u64 length; // the currently used # of entities in this chunk + u64 capacity; // the "chunk size" / space in this chunk + struct EntityChunk* next; // the next chunk + Entity* items; // the actual entities +} EntityChunk; + +typedef struct ChunkedEntityList { + u64 length; // the current number of entities + u64 chunk_size; // the # of entities per chunk + u64 chunks; // the # of chunks in this list so far + EntityChunk* first; // the first chunk of entities +} ChunkedEntityList; + +typedef struct Client { + u16 lan_port; + i32 lan_ip; + u64 character_eid; + u64 account_id; + SocketAddress address; + CommandType commands[CLIENT_COMMAND_LIST_LEN]; + u64 last_ping; +} Client; + +typedef struct ClientList { + u64 length; + u64 capacity; + Client* items; +} ClientList; + +typedef struct State { + Mutex client_mutex; + Mutex mutex; + ClientList clients; + u64 next_eid; + AccountChunk accounts; + u64 frame; + Arena game_scratch; + StringArena string_arena; + ParsedClientCommandThreadQueue* network_recv_queue; + OutgoingMessageQueue* network_send_queue; +} State; + +///// Global Variables +global State state = { 0 }; +global Arena permanent_arena = { 0 }; +global const Entity NULL_ENTITY = { 0 }; +global bool debug_mode = false; +global ChunkedEntityList free_chunks = { 0, CHUNK_SIZE, 0, NULL }; + +///// functionImplementations() +fn ParsedClientCommandThreadQueue* newPCCThreadQueue(Arena* a) { + ParsedClientCommandThreadQueue* result = arenaAlloc(a, sizeof(ParsedClientCommandThreadQueue)); + MemoryZero(result, (sizeof *result)); + result->mutex = newMutex(); + result->not_full = newCond(); + result->not_empty = newCond(); + return result; +} + +fn void pccThreadSafeQueuePush(ParsedClientCommandThreadQueue* queue, ParsedClientCommand* msg) { + lockMutex(&queue->mutex); { + while (queue->count == PARSED_CLIENT_COMMAND_THREAD_QUEUE_LEN) { + waitForCondSignal(&queue->not_full, &queue->mutex); + } + + MemoryCopy(&queue->items[queue->tail], msg, (sizeof *msg)); + queue->tail = (queue->tail + 1) % PARSED_CLIENT_COMMAND_THREAD_QUEUE_LEN; + queue->count++; + + signalCond(&queue->not_empty); + } unlockMutex(&queue->mutex); +} + +fn ParsedClientCommand* pccThreadSafeNonblockingQueuePop(ParsedClientCommandThreadQueue* q, ParsedClientCommand* copy_target) { + // immediately returns NULL if there's nothing in the ThreadQueue + // copies the ParsedClientCommand into `copy_target` if there is something in the queue + // and marks it as popped from the queue + ParsedClientCommand* result = NULL; + + lockMutex(&q->mutex); { + if (q->count > 0) { + result = &q->items[q->head]; + MemoryCopy(copy_target, result, (sizeof *copy_target)); + q->head = (q->head + 1) % PARSED_CLIENT_COMMAND_THREAD_QUEUE_LEN; + q->count--; + + signalCond(&q->not_full); + } + } unlockMutex(&q->mutex); + + return result; +} + +fn u64 entitySerialize(Entity current, Account* acct, u64 index, u8 bytes[]) { + // send entity header (common to all entity types) + index += writeU64ToBufferLE(bytes + index, current.id); + index += writeU64ToBufferLE(bytes + index, current.features); + bytes[index++] = current.x; + bytes[index++] = current.y; + bytes[index++] = (u8)current.type; // EntityType + + //if (CheckFlag(current.features, FeatureRegensHp)) { + // index += writeU16ToBufferLE(bytes + index, current.hp); + // index += writeU16ToBufferLE(bytes + index, current.max_hp); + //} + + // send character-specific details (color and name) + if (current.type == EntityCharacter) { + bytes[index++] = current.color; + // send name string + index += writeU64ToBufferLE(bytes + index, acct->name.length); + for (u32 j = 0; j < acct->name.length; j++) { + bytes[index++] = acct->name.bytes[j]; + } + } + return index; +} + +fn u64 entityFeaturesFromType(EntityType type) { + u64 result = 0; + switch (type) { + case EntityCharacter: { + SetFlag(result, FeatureWalksAround); + SetFlag(result, FeatureCanFight); + } break; + default: { + } break; + } + return result; +} + +fn Account* findAccountById(u64 id) { + AccountChunk* current = &state.accounts; + while (current != NULL) { + for (u32 i = 0; i < current->length; i++) { + if (current->items[i].id == id) { + return ¤t->items[i]; + } + } + current = current->next; + } + return NULL; +} + +fn Account* findAccountByEId(u64 id) { + AccountChunk* current = &state.accounts; + while (current != NULL) { + for (u32 i = 0; i < current->length; i++) { + if (current->items[i].eid == id) { + return ¤t->items[i]; + } + } + current = current->next; + } + return NULL; +} + +fn Account* findAccountByName(String name) { + AccountChunk* current = &state.accounts; + while (current != NULL) { + for (u32 i = 0; i < current->length; i++) { + if (stringsEq(¤t->items[i].name, &name)) { + return ¤t->items[i]; + } + } + current = current->next; + } + return NULL; +} + +fn Account* newAccount(Arena* a, Account details) { + u64 id = 0; + AccountChunk* current = &state.accounts; + while (current != NULL) { + id += current->length; + if (current->length < current->capacity) { + Account* result = ¤t->items[current->length]; + *result = details; + result->id = id; + current->length += 1; + printf("new account created id=%lld\n", id); + return result; + } + if (current->next == NULL) { + // alloc next chunk of accounts + AccountChunk* next = arenaAlloc(a, sizeof(AccountChunk)); + next->capacity = ACCOUNT_CHUNK_SIZE; + next->items = arenaAllocArray(a, Account, next->capacity); + current->next = next; + } + current = current->next; + } + assert(false); // never should be reached + return NULL; +} + +fn u64 pushEntity(EntityChunk* chunk, Entity e) { + assert(chunk->length < chunk->capacity); + chunk->items[chunk->length] = e; + u64 result = chunk->length; + chunk->length += 1; + return result; +} + +fn bool isSpaceInChunk(EntityChunk* chunk) { + return chunk->length < chunk->capacity; +} + +fn bool allChunksFull(ChunkedEntityList list) { + return list.length >= (list.chunk_size * list.chunks); +} + +fn EntityChunk* pushNewChunk(Arena* a, ChunkedEntityList* list) { + EntityChunk* new_chunk; + if (free_chunks.length > 0) { + new_chunk = free_chunks.first; + free_chunks.length -= 1; + free_chunks.first = new_chunk->next; + } else { + new_chunk = arenaAlloc(a, sizeof(EntityChunk)); + // alloc the new chunk of entities + new_chunk->length = 0; + new_chunk->capacity = list->chunk_size; + new_chunk->next = NULL; + new_chunk->items = arenaAllocArray(a, Entity, list->chunk_size); + } + // bookeeping in the list + list->chunks += 1; + if (list->first == NULL) { + list->first = new_chunk; + } else { + EntityChunk* last = list->first; + while (last->next != NULL) { + last = last->next; + } + last->next = new_chunk; + } + return new_chunk; +} + +fn Entity* entityPtrFromChunkList(ChunkedEntityList* list, i32 index) { + Entity* result = (Entity*)&NULL_ENTITY; + i32 chunk_index = index / list->chunk_size; + EntityChunk* chunk = list->first; + for (i32 i = 0; i < chunk_index; i++) { + chunk = chunk->next; + } + result = &chunk->items[index % list->chunk_size]; + return result; +} + +fn Entity entityFromChunkList(ChunkedEntityList* list, i32 index) { + Entity result = {0}; + i32 chunk_index = index / list->chunk_size; + EntityChunk* chunk = list->first; + for (i32 i = 0; i < chunk_index; i++) { + chunk = chunk->next; + } + result = chunk->items[index % list->chunk_size]; + return result; +} + +fn bool deleteLastEntity(ChunkedEntityList* list, EntityChunk* last_chunk, EntityChunk* second_to_last_chunk) { + last_chunk->length -= 1; + list->length -= 1; + // don't delete the room's only chunk, but otherwise move the chunk to the free-list + if (last_chunk->length == 0 && last_chunk != list->first) { + list->chunks -= 1; + second_to_last_chunk->next = NULL; + free_chunks.length += 1; + EntityChunk* last_free_chunk = free_chunks.first; + if (last_free_chunk) { + while (last_free_chunk->next != NULL) { + last_free_chunk = last_free_chunk->next; + } + last_free_chunk->next = last_chunk; + } else { + free_chunks.first = last_chunk; + } + } + return true; +} + +fn void exitWithErrorMessage(ptr msg) { + printf("error: %s", msg); + exit(1); +} + +fn u32 pushClient(ClientList* clients, SocketAddress addr) { + Client new_client = {0}; + new_client.last_ping = state.frame; + new_client.address = addr; + + // first, try to overwrite an old dc'ed client + for (u32 i = 1; i < clients->length; i++) { + if (clients->items[i].character_eid == 0) { + clients->items[i] = new_client; + return i; + } + } + + // if there are none, then add the client to the end of the list + assert(clients->capacity > clients->length);//TODO grow the list if we need to + clients->items[clients->length] = new_client; + u32 result = clients->length; + clients->length += 1; + return result; +} + +fn bool deleteClientByEId(ClientList* clients, u64 id) { + bool succeeded = false; + Client blank_client = {0}; + // i=1 because first client is null-client + for (u32 i = 1; i < clients->length && !succeeded; i++) { + if (clients->items[i].character_eid == id) { + clients->items[i] = blank_client; + succeeded = true; + } + } + return succeeded; +} + +fn u32 findClientHandleByEId(ClientList* clients, u64 id) { + for (u32 i = 0; i < clients->length; i++) { + Client c = clients->items[i]; + if (c.character_eid == id) { + return i; + } + } + return 0; +} + +fn u32 findClientHandleBySocketAddress(ClientList* clients, SocketAddress address) { + for (u32 i = 0; i < clients->length; i++) { + Client c = clients->items[i]; + if (socketAddressEqual(address, c.address)) { + return i; + } + } + return 0; +} + +fn void handleIncomingMessage(u8* message, u32 len, SocketAddress sender, i32 socket) { + dbg("%d: %s from %s:%d\n", len, command_type_strings[message[0]], inet_ntoa(sender.sin_addr), sender.sin_port); + u32 msg_idx = 0; + ParsedClientCommand parsed = { + .type = (CommandType)message[msg_idx++], + .sender_ip = sender.sin_addr.s_addr, + .sender_port = sender.sin_port, + }; + u8 temp_bytes[UDP_MAX_MESSAGE_LEN] = {0}; + String temp_str = { + .bytes = (char*)temp_bytes, + .length = 0, + .capacity = 0, + }; + switch (parsed.type) { + case CommandLogin: { + // parse the login command + parsed.alt_port = ~readU16FromBufferLE(message + msg_idx); + msg_idx += 2; + parsed.alt_ip = ~readI32FromBufferLE(message + msg_idx); + msg_idx += 4; + + // parse the name + u8 name_len = message[msg_idx++]; + MemoryZero(temp_bytes, UDP_MAX_MESSAGE_LEN); + temp_str.length = name_len; + temp_str.capacity = temp_str.length+1; + for (u32 j = 0; j < temp_str.length; j++) { + temp_str.bytes[j] = message[msg_idx+j]; + } + parsed.name = allocStringChunkList(&state.string_arena, temp_str); + msg_idx += temp_str.length; + + // parse the password + u32 pw_len = 0; + while (message[msg_idx+pw_len]) { + pw_len += 1; + } + MemoryZero(temp_bytes, UDP_MAX_MESSAGE_LEN); + temp_str.length = pw_len; + temp_str.capacity = temp_str.length+1; + for (u32 j = 0; j < temp_str.length; j++) { + temp_str.bytes[j] = message[msg_idx+j]; + } + parsed.pass = allocStringChunkList(&state.string_arena, temp_str); + msg_idx += temp_str.length; + + printf("Logging in player: %s %d %s\n", MESSAGE_STRINGS[parsed.type], name_len, message + 7); + } break; + case CommandKeepAlive: + break; + case CommandCreateCharacter: { + printf("command create character received\n"); + parsed.byte = message[1]; + } break; + case CommandInvalid: + case CommandType_Count: { + dbg("invalid command type"); + } break; + } + + pccThreadSafeQueuePush(state.network_recv_queue, &parsed); + fflush(stdout); +} + +fn void* receiveNetworkUpdates(void* udp) { + UDPServer server = *(UDPServer*)udp; + dbg("receiveNetworkUpdates() sock=%d\n", server.server_socket); + infiniteReadUDPServer(&server, handleIncomingMessage); + return NULL; +} + +fn void* sendNetworkUpdates(void* sock) { + ThreadContext tctx = {0}; + tctxInit(&tctx); + i32* socket_ptr = (i32*)sock; + i32 socket = *socket_ptr; + while (true) { + u64 loop_start = osTimeMicrosecondsNow(); + + // 1. clear out our "outgoingMessage" queue + { + UDPMessage to_send = { 0 }; + UDPMessage* next_to_send = outgoingMessageNonblockingQueuePop(state.network_send_queue, &to_send); + u8List bytes_list = { 0 }; + while (next_to_send != NULL) { + bytes_list.items = to_send.bytes; + bytes_list.length = to_send.bytes_len; + bytes_list.capacity = to_send.bytes_len; + sendUDPu8List(socket, &to_send.address, &bytes_list); + next_to_send = outgoingMessageNonblockingQueuePop(state.network_send_queue, &to_send); + } + } + + lockMutex(&state.client_mutex); { + // WARNING the `i` starts at 1 here because state.clients.items[0] is a "null" Client + for (u32 i = 1; i < state.clients.length; i++) { + Client client = state.clients.items[i]; + if (client.last_ping+CLIENT_TIMEOUT_FRAMES < state.frame) { + memset(&state.clients.items[i], 0, sizeof(Client)); + continue; + } + if (client.character_eid == 0) { + continue; // they are still creating their character + } + + } + } unlockMutex(&state.client_mutex); + + u32 loop_duration = osTimeMicrosecondsNow() - loop_start; + i32 remaining_time = GOAL_NETWORK_SEND_LOOP_US - loop_duration; + if (remaining_time > 0) { + osSleepMicroseconds(remaining_time); + } + } + return NULL; +} + +fn void* gameLoop(void* params) { + LaneCtx* lane_ctx = (LaneCtx*)params; + ThreadContext tctx = { + .lane_ctx = *lane_ctx, + }; + tctxInit(&tctx); + printf("Lane %lld (%lld) of %lld starting.\n", lane_ctx->lane_idx, LaneIdx(), lane_ctx->lane_count); + fflush(stdout); + UDPMessage outgoing_message = {0}; + u64 loop_start; + u64 last_burn = 0; + u64 last_hp_regen = 0; + Arena scratch_arena = {0}; + arenaInit(&scratch_arena); + while (true) { + loop_start = osTimeMicrosecondsNow(); + + if (LaneIdx() == 0) { // narrow + state.frame += 1; + + // 1. process client messages + lockMutex(&state.client_mutex); lockMutex(&state.mutex); { + + u32 msg_iters = 0; + SocketAddress sender = {0}; + ParsedClientCommand msg = {0}; + ParsedClientCommand* next_net_msg = pccThreadSafeNonblockingQueuePop(state.network_recv_queue, &msg); + while (next_net_msg != NULL) { + msg_iters += 0; + sender.sin_addr.s_addr = msg.sender_ip; + sender.sin_port = msg.sender_port; + // find which client it is + u32 client_handle = findClientHandleBySocketAddress(&state.clients, sender); + Client* client = &state.clients.items[client_handle]; + switch (msg.type) { + case CommandKeepAlive: { + dbg("KeepAlive for client_handle=%d, %ld", client_handle, state.frame); + state.clients.items[client_handle].last_ping = state.frame; + } break; + case CommandLogin: { + if (client_handle == 0) { + client_handle = pushClient(&state.clients, sender); + client = &state.clients.items[client_handle]; + printf("pushed new client handle = %d\n", client_handle); + } + // update/set the lan_ip/port info for p2p connections + client->lan_ip = htonl(msg.alt_ip); + client->lan_port = htons(msg.alt_port); + + /* + struct in_addr ipaddr; + ipaddr.s_addr = htonl(msg.alt_ip); + printf("client #%d: SENDER=%s:%d\n", client_handle, inet_ntoa(sender.sin_addr), ntohs(sender.sin_port)); + printf(" LAN=%s:%d %d vs %d vs %d\n", inet_ntoa(ipaddr), msg.alt_port, msg.alt_ip, htonl(msg.alt_ip), sender.sin_addr.s_addr); + */ + + String name = stringChunkToString(&permanent_arena, msg.name); + releaseStringChunkList(&state.string_arena, &msg.name); + + String pw = stringChunkToString(&permanent_arena, msg.pass); + releaseStringChunkList(&state.string_arena, &msg.pass); + + Account* existing_account = findAccountByName(name); + printf("name(%d): %s pw(%d): %s acct?: %d\n", name.length, name.bytes, pw.length, pw.bytes, existing_account != NULL); + fflush(stdout); + if (existing_account) { + printf(" existing account\n"); + bool pw_matches = stringsEq(&pw, &existing_account->pw); + arenaDealloc(&permanent_arena, pw.capacity); + arenaDealloc(&permanent_arena, name.capacity); + if (pw_matches) { + printf(" pw matched\n"); + } else { + // tell the client they did a bad pw + outgoing_message.bytes[0] = (u8)MessageBadPw; + outgoing_message.bytes_len = 1; + outgoing_message.address = sender; + outgoingMessageQueuePush(state.network_send_queue, &outgoing_message); + printf("MessageBadPw sent\n"); + break; + } + } else { + Account new_account = { + .eid = 0, + .name = name, + .pw = pw, + }; + existing_account = newAccount(&permanent_arena, new_account); + } + client->account_id = existing_account->id; + if (existing_account->eid != 0) { + client->character_eid = existing_account->eid; + + // tell the client their character id + outgoing_message.bytes[0] = (u8)MessageCharacterId; + writeU64ToBufferLE(outgoing_message.bytes + 1, existing_account->eid); + outgoing_message.bytes_len = 9; + outgoing_message.address = sender; + outgoingMessageQueuePush(state.network_send_queue, &outgoing_message); + printf("MessageCharacterId sent\n"); + } else { + // tell the client they made a new account + outgoing_message.bytes[0] = (u8)MessageNewAccountCreated; + outgoing_message.bytes_len = 1; + outgoing_message.address = sender; + outgoingMessageQueuePush(state.network_send_queue, &outgoing_message); + printf("MessageNewAccountCreated sent\n"); + } + printf("eid=%lld, client_handle=%d, acct_id=%lld\n", existing_account->eid, client_handle, existing_account->id); + } break; + case CommandCreateCharacter: { + if (client->character_eid == 0) { + // Create new character + XYZ room_xyz = {0, 0, 0 }; + Entity character = { + .type = EntityCharacter, + .id = state.next_eid++, + .changed = true, + .features = entityFeaturesFromType(EntityCharacter), + .color = msg.byte, + }; + dbg("made new character id=%ld\n", character.id); + client->character_eid = character.id; + Account* account = findAccountById(client->account_id); + account->eid = character.id; + printf("character_eid=%lld, client_handle=%d, acct_id=%lld\n", account->eid, client_handle, account->id); + + // tell the client their character id + outgoing_message.address = sender; + outgoing_message.bytes_len = 9; + outgoing_message.bytes[0] = (u8)MessageCharacterId; + writeU64ToBufferLE(outgoing_message.bytes + 1, character.id); + outgoingMessageQueuePush(state.network_send_queue, &outgoing_message); + printf("MessageCharacterId sent\n"); + } else { + printf("client tried to create a character when he already has one."); + } + } break; + case CommandType_Count: + case CommandInvalid: + dbg("invalid message from queue"); + break; + } + next_net_msg = pccThreadSafeNonblockingQueuePop(state.network_recv_queue, &msg); + msg_iters++; + } + + } unlockMutex(&state.mutex); unlockMutex(&state.client_mutex); + } + + LaneSync(); + + // 2. tick non-user entities + // iterate all the rooms + /* + Room* room = NULL; + Range1u64 room_range = LaneRange(MAX_ROOMS); + for (u32 i = room_range.min; i < room_range.max; i++) { + room = &state.rooms->items[i]; + simulateRoom(&scratch_arena, room, burn_tick, regen_hp_tick); + } + */ + + // 3. scratch cleanup + arenaClear(&scratch_arena); + + // 4. loop timing + u32 loop_duration = osTimeMicrosecondsNow() - loop_start; + i32 remaining_time = GOAL_GAME_LOOP_US - loop_duration; + if (remaining_time > 0) { + osSleepMicroseconds(remaining_time); + } + } + return NULL; +} + +// THE SERVER +i32 main(i32 argc, ptr argv[]) { + osInit(); + // multi-thread architecture: + // - the gameLoop() which just inifinite loops every "tick" and processes user input and updates gameworld state + // - TODO: actually make this N "lanes" as described https://www.rfleury.com/p/multi-core-by-default + // such that we can split room-work among all the various cores on our machine + // - one is the sendNetworkUpdates() infinite loop, which sends a UDP snapshot-or-delta update to each connected client N/sec + // - one is the receiveNetworkUpdates() infinte loop, which waits for new UDP messages from clients + + // 1. initialize gameworld, and spin off infinite game-loop thread + // 2. spin off sendNetworkUpdates() infinite loop thread + // 3. infinitely wait for incoming UDP messages and process them (usually by just dropping user-commands into the relevant block of shared memory) + + // 1. initialize gameworld, and spin off infinite game-loop thread + arenaInit(&permanent_arena); + arenaInit(&state.game_scratch); + arenaInit(&state.string_arena.a); + state.string_arena.mutex = newMutex(); + state.client_mutex = newMutex(); + state.mutex = newMutex(); + state.network_recv_queue = newPCCThreadQueue(&permanent_arena); + state.network_send_queue = newOutgoingMessageQueue(&permanent_arena); + // alloc the global hashmap of rooms + // init + alloc clients + state.clients.capacity = SERVER_MAX_CLIENTS; + state.clients.length = 1; // making entry 0 to be a "null" client + state.clients.items = (Client*)arenaAllocArray(&permanent_arena, Client, SERVER_MAX_CLIENTS); + state.accounts.capacity = ACCOUNT_CHUNK_SIZE; + state.accounts.items = arenaAllocArray(&permanent_arena, Account, ACCOUNT_CHUNK_SIZE); + + // 2. spin off sendNetworkUpdates() infinite loop thread + UDPServer listener = createUDPServer(SERVER_PORT); + if (!listener.ready) { + exitWithErrorMessage("Couldn't start the udp server"); + } + Thread send_thread = spawnThread(&sendNetworkUpdates, &listener.server_socket); + // 3. infinitely wait for incoming UDP messages and process them (usually by just dropping user-commands into the relevant block of shared memory) + Thread recv_thread = spawnThread(&receiveNetworkUpdates, &listener); + + u64 lane_broadcast_val = 0; + Barrier barrier = osBarrierAlloc(GAME_THREAD_CONCURRENCY); + LaneCtx lane_ctxs[GAME_THREAD_CONCURRENCY] = {0}; + Thread game_threads[GAME_THREAD_CONCURRENCY] = {0}; + for (u32 i = 0; i < GAME_THREAD_CONCURRENCY; i++) { + lane_ctxs[i].lane_idx = i; + lane_ctxs[i].lane_count = GAME_THREAD_CONCURRENCY; + lane_ctxs[i].barrier = barrier; + lane_ctxs[i].broadcast_memory = &lane_broadcast_val; + game_threads[i] = spawnThread(&gameLoop, &lane_ctxs[i]); + } + for (u32 i = 0; i < GAME_THREAD_CONCURRENCY; i++) { + osThreadJoin(game_threads[i], MAX_u64); + } + + return 0; +} + diff --git a/src/shared.h b/src/shared.h new file mode 100644 index 0000000..e2bb99f --- /dev/null +++ b/src/shared.h @@ -0,0 +1,77 @@ +#include "base/all.h" + +#ifndef GAMESHARED_H +#define GAMESHARED_H + +///// #define some game-tunable constants +#define GAME_CONSTANT_ONE (1) +#define GAME_CONSTANT_TWO (2) + +typedef enum EntityFeature { + FeatureWalksAround, + FeatureCanFight, + EntityFeature_Count +} EntityFeature; + +typedef enum EntityType { + EntityNull, + EntityWall, + EntityDoor, + EntityCharacter, + EntityType_Count, +} EntityType; +static const char* ENTITY_STRINGS[] = { + "NULL", + "Wall", + "Door", + "Character", +}; + +typedef struct { + i32 x; + i32 y; + i32 z; +} XYZ; + +typedef enum Direction { + DirectionInvalid, + North, + South, + East, + West, + Up, + Down, + Direction_Count +} Direction; + +typedef enum CommandType { + CommandInvalid, + CommandKeepAlive, + CommandLogin, + CommandCreateCharacter, + CommandType_Count, +} CommandType; +static const char* command_type_strings[] = { + "Invalid", + "KeepAlive", + "Login", + "CreateCharacter", +}; + +#define ENTITY_HEADER_MESSAGE_SIZE (8+8+1+1+1) +#define ENTITY_MESSAGE_SIZE (ENTITY_HEADER_MESSAGE_SIZE+2+2+2+2+8+1+1) +typedef enum Message { + MessageInvalid, + MessageCharacterId, + MessageBadPw, + MessageNewAccountCreated, + Message_Count, +} Message; +static const char* MESSAGE_STRINGS[] = { + "Invalid", + "CharacterId", + "BadPw", + "NewAccountCreated", +}; + +#endif //GAMESHARED_H diff --git a/src/string_chunk.c b/src/string_chunk.c new file mode 100644 index 0000000..5e83cbf --- /dev/null +++ b/src/string_chunk.c @@ -0,0 +1,162 @@ +#include "string_chunk.h" + +fn StringChunkList allocStringChunkList(StringArena* a, String string) { + StringChunkList result = {0}; + u64 needed_chunks = (string.length + (STRING_CHUNK_PAYLOAD_SIZE-1)) / STRING_CHUNK_PAYLOAD_SIZE; + u64 bytes_left = string.length; + u64 string_offset = 0; + lockMutex(&a->mutex); { + for (u32 i = 0; i < needed_chunks; i++) { + StringChunk* chunk = a->first_free_str_chunk; + if (chunk == NULL) { + chunk = (StringChunk*)arenaAlloc(&a->a, sizeof(StringChunk)+STRING_CHUNK_PAYLOAD_SIZE); + } else { + a->first_free_str_chunk = a->first_free_str_chunk->next; + } + chunk->next = NULL; // makes sure we don't have a pointer to any other free_str_chunks + u64 bytes_to_copy = Min(bytes_left, STRING_CHUNK_PAYLOAD_SIZE); + // ryan's impl used chunk+1 which seems like a bug but what do I know he had a working demo + MemoryCopy(chunk+1, string.bytes+string_offset, bytes_to_copy); + QueuePush(result.first, result.last, chunk); + result.count += 1; + result.total_size += bytes_to_copy; + bytes_left -= bytes_to_copy; + string_offset += bytes_to_copy; + } + } unlockMutex(&a->mutex); + return result; +} + +fn void releaseStringChunkList(StringArena* a, StringChunkList* list) { + StringChunk* chunk = list->first; + lockMutex(&a->mutex); { + for (StringChunk* next = NULL; chunk != NULL; chunk = next) { + next = chunk->next; + chunk->next = a->first_free_str_chunk; + a->first_free_str_chunk = chunk; + } + } unlockMutex(&a->mutex); + MemoryZeroStruct(list, StringChunkList); +} + +fn String stringChunkToString(Arena* a, StringChunkList list) { + String result = { + .length = list.total_size, + .capacity = list.total_size + 1, + .bytes = arenaAllocArray(a, u8, list.total_size+1), + }; + // copy the string bytes out of the StringChunkList into the correctly-sized String + StringChunk* chunk = list.first; + for (u32 i = 0; i < list.total_size; i++) { + if (i > 0 && i % STRING_CHUNK_PAYLOAD_SIZE == 0) { + chunk = chunk->next; + } + result.bytes[i] = *((char*)(chunk + 1) + (i%STRING_CHUNK_PAYLOAD_SIZE)); + } + return result; +} + +fn void stringChunkListAppend(StringArena* a, StringChunkList* list, String string) { + u64 capacity = list->count * STRING_CHUNK_PAYLOAD_SIZE; + u64 remaining_space = capacity - list->total_size; + u64 bytes_in_last_chunk = (list->total_size % STRING_CHUNK_PAYLOAD_SIZE); + if (string.length < remaining_space) { + MemoryCopy(((u8*)list->last)+sizeof(StringChunk*)+(bytes_in_last_chunk), string.bytes, string.length); + list->total_size += string.length; + } else { + u64 string_offset = 0; + u64 bytes_left = string.length; + u64 bytes_to_copy = Min(bytes_left, remaining_space); + if (remaining_space > 0) { + // fill up the last chunk + MemoryCopy(((u8*)(list->last+1))+(bytes_in_last_chunk), string.bytes, bytes_to_copy); + bytes_left -= bytes_to_copy; + string_offset += bytes_to_copy; + list->total_size += bytes_to_copy; + } + + // then figure out how many more chunks we need + u64 needed_chunks = (bytes_left + (STRING_CHUNK_PAYLOAD_SIZE-1)) / STRING_CHUNK_PAYLOAD_SIZE; + lockMutex(&a->mutex); { + for (u32 i = 0; i < needed_chunks; i++) { + StringChunk* chunk = a->first_free_str_chunk; + if (chunk == NULL) { + chunk = (StringChunk*)arenaAlloc(&a->a, sizeof(StringChunk)+STRING_CHUNK_PAYLOAD_SIZE); + } else { + a->first_free_str_chunk = a->first_free_str_chunk->next; + } + chunk->next = NULL; // makes sure we don't have a pointer to any other free_str_chunks + bytes_to_copy = Min(bytes_left, STRING_CHUNK_PAYLOAD_SIZE); + // ryan's impl used chunk+1 which seems like a bug but what do I know he had a working demo + MemoryCopy(chunk+1, string.bytes+string_offset, bytes_to_copy); + QueuePush(list->first, list->last, chunk); + list->count += 1; + list->total_size += bytes_to_copy; + bytes_left -= bytes_to_copy; + string_offset += bytes_to_copy; + } + } unlockMutex(&a->mutex); + } +} + +fn void stringChunkListDeleteLast(StringArena* a, StringChunkList* list) { + if (list->total_size == 0) return; + + u64 capacity = list->count * STRING_CHUNK_PAYLOAD_SIZE; + u64 remaining_space = capacity - list->total_size; + u64 bytes_in_last_chunk = STRING_CHUNK_PAYLOAD_SIZE - remaining_space; + + bool theres_only_one_chunk = list->total_size <= STRING_CHUNK_PAYLOAD_SIZE; + bool last_chunk_only_has_one_byte = bytes_in_last_chunk == 1; + + if (theres_only_one_chunk) { + ptr char_to_delete = ((char*)(list->last + 1) + (list->total_size-1)); + char_to_delete[0] = '\0'; + } else if (last_chunk_only_has_one_byte) { + // remove the last chunk when it's not the only chunk + StringChunk* second_to_last_chunk = list->first; + while (second_to_last_chunk->next != list->last && second_to_last_chunk->next != NULL) { + second_to_last_chunk = second_to_last_chunk->next; + } + second_to_last_chunk->next = NULL; + lockMutex(&a->mutex); { + list->last->next = a->first_free_str_chunk; + a->first_free_str_chunk = list->last; + } unlockMutex(&a->mutex); + list->last = second_to_last_chunk; + list->count -= 1; + } else { + ptr char_to_delete = ((char*)(list->last + 1) + ((list->total_size-1) % STRING_CHUNK_PAYLOAD_SIZE)); + char_to_delete[0] = '\0'; + } + list->total_size -= 1; +} + +fn StringChunkList stringChunkListInit(StringArena* a) { + StringChunkList result = {0}; + lockMutex(&a->mutex); { + StringChunk* chunk = a->first_free_str_chunk; + if (chunk == NULL) { + chunk = (StringChunk*)arenaAlloc(&a->a, sizeof(StringChunk)+STRING_CHUNK_PAYLOAD_SIZE); + } else { + a->first_free_str_chunk = a->first_free_str_chunk->next; + } + chunk->next = NULL; // makes sure we don't have a pointer to any other free_str_chunks + MemoryZero(chunk+1, STRING_CHUNK_PAYLOAD_SIZE); + QueuePush(result.first, result.last, chunk); + result.count += 1; + } unlockMutex(&a->mutex); + return result; +} + +fn void stringChunkCopyToBuffer(StringChunkList* list, u8* buffer, u32 len) { + assert(list->total_size <= len); + + StringChunk* chunk = list->first; + for (u32 i = 0; i < list->total_size; i++) { + if (i > 0 && i % STRING_CHUNK_PAYLOAD_SIZE == 0) { + chunk = chunk->next; + } + buffer[i] = *((char*)(chunk + 1) + (i%STRING_CHUNK_PAYLOAD_SIZE)); + } +} diff --git a/src/string_chunk.h b/src/string_chunk.h new file mode 100644 index 0000000..a9c0baf --- /dev/null +++ b/src/string_chunk.h @@ -0,0 +1,33 @@ +#ifndef STRING_CHUNK_H +#define STRING_CHUNK_H + +#include "base/all.h" + +#define STRING_CHUNK_PAYLOAD_SIZE (64 - sizeof(StringChunk*)) + +typedef struct StringChunk { + struct StringChunk *next; // essentially a header, followed by a fixed maximum str bytes +} StringChunk; + +typedef struct StringChunkList { + StringChunk* first; + StringChunk* last; + u64 count; + u64 total_size; +} StringChunkList; + +typedef struct StringArena { + Arena a; + StringChunk* first_free_str_chunk; + Mutex mutex; +} StringArena; + +fn StringChunkList allocStringChunkList(StringArena* a, String string); +fn void releaseStringChunkList(StringArena* a, StringChunkList* list); +fn String stringChunkToString(Arena* a, StringChunkList list); +fn void stringChunkListAppend(StringArena* a, StringChunkList* list, String string); +fn void stringChunkListDeleteLast(StringArena* a, StringChunkList* list); +fn StringChunkList stringChunkListInit(StringArena* a); +fn void stringChunkCopyToBuffer(StringChunkList* list, u8* buffer, u32 len); + +#endif //STRING_CHUNK_H diff --git a/win_make.bat b/win_make.bat new file mode 100644 index 0000000..20256f0 --- /dev/null +++ b/win_make.bat @@ -0,0 +1,4 @@ +@echo off + +del /q .\build\client.exe 2>nul +gcc --static -std=c99 -g -o build/client src/client.c -lpthread -lws2_32 -lwinmm -liphlpapi