fsound.h

swf文件结构介绍

#ifndef SNDMIX_INCLUDED
#define SNDMIX_INCLUDED

// #include <windows.h>
// #include <mmreg.h>
#include <assert.h>
#include <stdio.h>
#include <vector>

#include "Macromedia.h"
#include "FFixed.h"

struct WaveFormat
{
    U16        wFormatTag;         /* format type */
    U16        nChannels;          /* number of channels (i.e. mono, stereo...) */
    U32        nSamplesPerSec;     /* sample rate */
    U32        nAvgBytesPerSec;    /* for buffer estimation */
    U16        nBlockAlign;        /* block size of data */
    U16        wBitsPerSample;     /* number of bits per sample of mono data */
    U16        cbSize;             /* the count in bytes of the size of */
};

// Our supported sample rates
// Note that sndRate5K is really 5512.5khz
// Use defines instead of an enum since these must be 32 bits on all platforms
#define sndRate5K	5512L
#define sndRate11K	11025L
#define sndRate22K	22050L
#define sndRate44K	44100L

const int sndMono			= 0x0;
const int sndStereo			= 0x1;

const int snd8Bit			= 0x0;
const int snd16Bit			= 0x2;

const int snd5K				= 0<<2;
const int snd11K			= 1<<2;
const int snd22K			= 2<<2;
const int snd44K			= 3<<2;

const int sndCompressNone	= 0x00; // we could add 14 more compression types here...
const int sndCompressADPCM	= 0x10;
const int sndCompressMP3	= 0x20;
const int sndCompressNoneI	= 0x30; // save out in intel byte order
const int sndRateMask		= 0x3 << 2;
const int sndCompressMask	= 0xF0;


enum {// Sound format types
		snd5K8Mono = 0,		snd5K8Stereo, 
		snd5K16Mono,		snd5K16Stereo, 
		snd11K8Mono,		snd11K8Stereo, 
		snd11K16Mono,		snd11K16Stereo, 
		snd22K8Mono,		snd22K8Stereo, 
		snd22K16Mono,		snd22K16Stereo, 
		snd44K8Mono,		snd44K8Stereo, 
		snd44K16Mono,		snd44K16Stereo 
	  };

// This object defines a sound sample
class FSound {
public:
	int			format;		
	S32			nSamples; 	// the number of samples - duration = nSamples/Rate()
	void *	samples;  	// this should probably be a handle on Mac
	S32			dataLen;   	// length in bytes of samples - set only if needed
    S32         delay;      // MP3 compression has a delay before the real sound data

	static const S32 kRateTable[4];
	static const int kRateShiftTable[4];
	
public:
	void Init();

	S32 Rate()				{ return kRateTable[(format >> 2)&0x3]; }
	int RateShift()			{ return kRateShiftTable[(format >> 2)&0x3]; }
	bool Stereo()			{ return (format & sndStereo) != 0; }
	int NChannels()			{ return (format & sndStereo) ? 2 : 1; }
	bool Is8Bit()			{ return (format & snd16Bit) == 0; }
	int BitsPerSample()		{ return (format & snd16Bit) ? 16 : 8; }
	int BytesPerSample()	{ return (format & snd16Bit) ? 2 : 1; }
	int CompressFormat()	{ return format & sndCompressMask; }
	bool Compressed()		{ return (format & sndCompressMask) != 0; }

	// Manage the duration in 44kHz units
	S32 GetDuration44() { return nSamples<<RateShift(); }
	void SetDuration44(S32 d) { nSamples = d>>RateShift(); }

	int BytesPerBlock() { return BytesPerSample()*NChannels(); }
	S32 SizeBytes() { return nSamples*BytesPerBlock(); }

	void Set( WaveFormat* );
};

class FSoundComp 
{
private:
	BOOL isStereo;
	BOOL is8Bit;
	int nBits;			// number of bits in each sample

	S32 nSamples;		// samples compressed so far

	S32 valpred[2];		// ADPCM state
	int index[2];

	// The Destination
	S32 len;			// default is to just calculate the size

	// State for writing bits
	unsigned int bitBuf;
	int bitPos;

  public:
	FSoundComp(FSound* snd, S32 numberBits );  // numberBits from 2-5
	~FSoundComp() { assert(bitPos == 0); }

	void Compress(void* src, S32 n, std::vector<U8>* stream);
	void Flush( std::vector<U8>* stream );

  private:
	void Compress16(S16* src, S32 n, std::vector<U8>* stream );

	// Write variable width bit fields
	void WriteBits( std::vector<U8>* stream );	// empty the buffer of whole bytes
	
	void PutBits(S32 v, int n, std::vector<U8>* stream ) {
		assert(n<=16);
		if ( bitPos + n > 32 ) WriteBits( stream );
		bitBuf = (bitBuf<<n) | (v & ~(0xFFFFFFFFL<<n));
		bitPos += n;
	}
};

#endif