注：此前已经写了一系列分析MediaInfo源代码的文章，列表如下：
MediaInfo源代码分析 1：整体结构
MediaInfo源代码分析 2：API函数
MediaInfo源代码分析 3：Open()函数
MediaInfo源代码分析 4：Inform()函数
MediaInfo源代码分析 5：JPEG解析代码分析
===================
本文分析MediaInfo中解码JPEG信息的模块。之前写了几篇文章都是关于MediaInfo主程序的，并没有分析其具体是如何解析不同多媒体文件信息的。在这里分析一下解码JPEG文件的代码。其他格式如BMP，GIF等解析的思路基本上是类似的。
File_Jpeg.h的File_Jpeg类的定义如下所示：
//***************************************************************************
// Class File_Jpeg
//***************************************************************************
//继承 File__Analyze
class File_Jpeg : public File__Analyze
{
public :
//In
stream_t StreamKind;
bool Interlaced;
//Constructor/Destructor
File_Jpeg();
private :
//Streams management
void Streams_Accept();
//Buffer - File header
bool FileHeader_Begin();
//Buffer - Synchro
bool Synchronize();
bool Synched_Test();
void Synched_Init();
//Buffer - Demux
#if MEDIAINFO_DEMUX
bool Demux_UnpacketizeContainer_Test() {return Demux_UnpacketizeContainer_Test_OneFramePerFile();}
#endif //MEDIAINFO_DEMUX
//Buffer - Global
void Read_Buffer_Unsynched();
#if MEDIAINFO_SEEK
size_t Read_Buffer_Seek (size_t Method, int64u Value, int64u ID) {return Read_Buffer_Seek_OneFramePerFile(Method, Value, ID);}
#endif //MEDIAINFO_SEEK
//Buffer - Per element
//解析头
void Header_Parse();
bool Header_Parser_Fill_Size();
//解析数据
void Data_Parse();
//Elements
//JPEG中的单元
//解析相应的单元，并获得信息
void TEM () {};
void SOC () {}
void SIZ ();
void COD ();
void COC () {Skip_XX(Element_Size, "Data");}
void TLM () {Skip_XX(Element_Size, "Data");}
void PLM () {Skip_XX(Element_Size, "Data");}
void PLT () {Skip_XX(Element_Size, "Data");}
void QCD ();
void QCC () {Skip_XX(Element_Size, "Data");}
void RGN () {Skip_XX(Element_Size, "Data");}
void PPM () {Skip_XX(Element_Size, "Data");}
void PPT () {Skip_XX(Element_Size, "Data");}
void CME () {Skip_XX(Element_Size, "Data");}
void SOT () {Skip_XX(Element_Size, "Data");}
void SOP () {Skip_XX(Element_Size, "Data");}
void EPH () {Skip_XX(Element_Size, "Data");}
void SOD ();
void SOF_();
void S0F0() {SOF_();};
void S0F1() {SOF_();};
void S0F2() {SOF_();};
void S0F3() {SOF_();}
void DHT () {Skip_XX(Element_Size, "Data");}
void S0F5() {SOF_();}
void S0F6() {SOF_();}
void S0F7() {SOF_();}
void JPG () {Skip_XX(Element_Size, "Data");}
void S0F9() {SOF_();}
void S0FA() {SOF_();}
void S0FB() {SOF_();}
void DAC () {Skip_XX(Element_Size, "Data");}
void S0FD() {SOF_();}
void S0FE() {SOF_();}
void S0FF() {SOF_();}
void RST0() {};
void RST1() {};
void RST2() {};
void RST3() {};
void RST4() {};
void RST5() {};
void RST6() {};
void RST7() {};
void SOI () {};
void EOI () {};
void SOS ();
void DQT () {Skip_XX(Element_Size, "Data");}
void DNL () {Skip_XX(Element_Size, "Data");}
void DRI () {Skip_XX(Element_Size, "Data");}
void DHP () {Skip_XX(Element_Size, "Data");}
void EXP () {Skip_XX(Element_Size, "Data");}
void APP0();
void APP0_AVI1();
void APP0_JFIF();
void APP0_JFFF();
void APP0_JFFF_JPEG();
void APP0_JFFF_1B();
void APP0_JFFF_3B();
void APP1();
void APP1_EXIF();
void APP2() {Skip_XX(Element_Size, "Data");}
void APP3() {Skip_XX(Element_Size, "Data");}
void APP4() {Skip_XX(Element_Size, "Data");}
void APP5() {Skip_XX(Element_Size, "Data");}
void APP6() {Skip_XX(Element_Size, "Data");}
void APP7() {Skip_XX(Element_Size, "Data");}
void APP8() {Skip_XX(Element_Size, "Data");}
void APP9() {Skip_XX(Element_Size, "Data");}
void APPA() {Skip_XX(Element_Size, "Data");}
void APPB() {Skip_XX(Element_Size, "Data");}
void APPC() {Skip_XX(Element_Size, "Data");}
void APPD() {Skip_XX(Element_Size, "Data");}
void APPE();
void APPE_Adobe0();
void APPF() {Skip_XX(Element_Size, "Data");}
void JPG0() {Skip_XX(Element_Size, "Data");}
void JPG1() {Skip_XX(Element_Size, "Data");}
void JPG2() {Skip_XX(Element_Size, "Data");}
void JPG3() {Skip_XX(Element_Size, "Data");}
void JPG4() {Skip_XX(Element_Size, "Data");}
void JPG5() {Skip_XX(Element_Size, "Data");}
void JPG6() {Skip_XX(Element_Size, "Data");}
void JPG7() {Skip_XX(Element_Size, "Data");}
void JPG8() {Skip_XX(Element_Size, "Data");}
void JPG9() {Skip_XX(Element_Size, "Data");}
void JPGA() {Skip_XX(Element_Size, "Data");}
void JPGB() {Skip_XX(Element_Size, "Data");}
void JPGC() {Skip_XX(Element_Size, "Data");}
void JPGD() {Skip_XX(Element_Size, "Data");}
void COM () {Skip_XX(Element_Size, "Data");}
//Temp
int8u APPE_Adobe0_transform;
bool APP0_JFIF_Parsed;
bool SOS_SOD_Parsed;
};
上面代码有以下几个特点：
1.继承了File__Analyze类
2.包含了很多JPEG中的数据单元的解析：DHT()，DQT()等等
下面来分别仔细看看源代码：
1.File__Analyze类代码巨多无比，先不分析。他继承了继承了File__Base
2.看一个解码具体单元的代码：SOF_()
注：SOF0（Start of Image，图像开始）。
SOF0，Start of Frame，帧图像开始
标记代码 2字节 固定值0xFFC0
包含9个具体字段：
① 数据长度 2字节 ①~⑥六个字段的总长度
即不包括标记代码，但包括本字段
② 精度 1字节 每个数据样本的位数
通常是8位，一般软件都不支持 12位和16位
③ 图像高度 2字节 图像高度（单位：像素），如果不支持 DNL 就必须 >0
④ 图像宽度 2字节 图像宽度（单位：像素），如果不支持 DNL 就必须 >0
⑤ 颜色分量数 1字节 只有3个数值可选
1：灰度图；3：YCrCb或YIQ；4：CMYK
而JFIF中使用YCrCb，故这里颜色分量数恒为3
⑥颜色分量信息 颜色分量数×3字节（通常为9字节）
a) 颜色分量ID 1字节
b) 水平/垂直采样因子 1字节 高4位：水平采样因子
低4位：垂直采样因子
（曾经看到某资料把这两者调转了）
c) 量化表 1字节 当前分量使用的量化表的ID
本标记段中，字段⑥应该重复出现，有多少个颜色分量（字段⑤），就出现多少次（一般为3次）。
=====================================
void File_Jpeg::SOF_()
{
//Parsing
vector<Jpeg_samplingfactor> SamplingFactors;
int16u Height, Width;
int8u Resolution, Count;
Get_B1 (Resolution, "P - Sample precision");
Get_B2 (Height, "Y - Number of lines");
Get_B2 (Width, "X - Number of samples per line");
Get_B1 (Count, "Nf - Number of image components in frame");
for (int8u Pos=0; Pos<Count; Pos++)
{
Jpeg_samplingfactor SamplingFactor;
Element_Begin1("Component");
Get_B1 ( SamplingFactor.Ci, "Ci - Component identifier"); if (SamplingFactor.Ci>Count) Element_Info1(Ztring().append(1, (Char)SamplingFactor.Ci)); else Element_Info1(SamplingFactor.Ci);
BS_Begin();
Get_S1 (4, SamplingFactor.Hi, "Hi - Horizontal sampling factor"); Element_Info1(SamplingFactor.Hi);
Get_S1 (4, SamplingFactor.Vi, "Vi - Vertical sampling factor"); Element_Info1(SamplingFactor.Vi);
BS_End();
Skip_B1( "Tqi - Quantization table destination selector");
Element_End0();
//Filling list of HiVi
SamplingFactors.push_back(SamplingFactor);
}
FILLING_BEGIN_PRECISE();
if (Frame_Count==0 && Field_Count==0)
{
Accept("JPEG");
Fill("JPEG");
if (Count_Get(StreamKind_Last)==0)
Stream_Prepare(StreamKind_Last);
Fill(StreamKind_Last, 0, Fill_Parameter(StreamKind_Last, Generic_Format), "JPEG");
Fill(StreamKind_Last, 0, Fill_Parameter(StreamKind_Last, Generic_Codec), "JPEG");
if (StreamKind_Last==Stream_Image)
Fill(Stream_Image, 0, Image_Codec_String, "JPEG", Unlimited, true, true); //To Avoid automatic filling
if (StreamKind_Last==Stream_Video)
Fill(Stream_Video, 0, Video_InternetMediaType, "video/JPEG", Unlimited, true, true);
Fill(StreamKind_Last, 0, Fill_Parameter(StreamKind_Last, Generic_BitDepth), Resolution);
Fill(StreamKind_Last, 0, "Height", Height*(Interlaced?2:1));
Fill(StreamKind_Last, 0, "Width", Width);
//ColorSpace from http://docs.oracle.com/javase/1.4.2/docs/api/javax/imageio/metadata/doc-files/jpeg_metadata.html
switch (APPE_Adobe0_transform)
{
case 0x01 :
if (Count==3)
Fill(StreamKind_Last, 0, "ColorSpace", "YUV");
case 0x02 :
if (Count==4)
Fill(StreamKind_Last, 0, "ColorSpace", "YCCB");
break;
default :
{
int8u Ci[256];
memset(Ci, 0, 256);;
for (int8u Pos=0; Pos<Count; Pos++)
Ci[SamplingFactors[Pos].Ci]++;
switch (Count)
{
case 1 : Fill(StreamKind_Last, 0, "ColorSpace", "Y"); break;
case 2 : Fill(StreamKind_Last, 0, "ColorSpace", "YA"); break;
case 3 :
if (!APP0_JFIF_Parsed && Ci['R']==1 && Ci['G']==1 && Ci['B']==1) //RGB
Fill(StreamKind_Last, 0, "ColorSpace", "RGB");
else if ((Ci['Y']==1 && ((Ci['C']==1 && Ci['c']==1) //YCc
|| Ci['C']==2)) //YCC
|| APP0_JFIF_Parsed //APP0 JFIF header present so YCC
|| APPE_Adobe0_transform==0 //transform set to YCC
|| (SamplingFactors[0].Ci==0 && SamplingFactors[1].Ci==1 && SamplingFactors[2].Ci==2) //012
|| (SamplingFactors[0].Ci==1 && SamplingFactors[1].Ci==2 && SamplingFactors[2].Ci==3)) //123
Fill(StreamKind_Last, 0, "ColorSpace", "YUV");
break;
case 4 :
if (!APP0_JFIF_Parsed && Ci['R']==1 && Ci['G']==1 && Ci['B']==1 && Ci['A']==1) //RGBA
Fill(StreamKind_Last, 0, "ColorSpace", "RGBA");
else if ((Ci['Y']==1 && Ci['A']==1 && ((Ci['C']==1 && Ci['c']==1) //YCcA
|| Ci['C']==2)) //YCCA
|| APP0_JFIF_Parsed //APP0 JFIF header present so YCCA
|| (SamplingFactors[0].Ci==0 && SamplingFactors[1].Ci==1 && SamplingFactors[2].Ci==2 && SamplingFactors[3].Ci==3) //0123
|| (SamplingFactors[0].Ci==1 && SamplingFactors[1].Ci==2 && SamplingFactors[2].Ci==3 && SamplingFactors[3].Ci==4)) //1234
Fill(StreamKind_Last, 0, "ColorSpace", "YUVA");
else if (APPE_Adobe0_transform==0) //transform set to CMYK
Fill(StreamKind_Last, 0, "ColorSpace", "YCCB");
break;
default: ;
}
}
}
//Chroma subsampling
if ((SamplingFactors.size()==3 || SamplingFactors.size()==4) && SamplingFactors[1].Hi==1 && SamplingFactors[2].Hi==1 && SamplingFactors[1].Vi==1 && SamplingFactors[2].Vi==1)
{
string ChromaSubsampling;
switch (SamplingFactors[0].Hi)
{
case 1 :
switch (SamplingFactors[0].Vi)
{
case 1 : ChromaSubsampling="4:4:4"; break;
default: ;
}
break;
case 2 :
switch (SamplingFactors[0].Vi)
{
case 1 : ChromaSubsampling="4:2:2"; break;
case 2 : ChromaSubsampling="4:2:0"; break;
default: ;
}
break;
case 4 :
switch (SamplingFactors[0].Vi)
{
case 1 : ChromaSubsampling="4:1:1"; break;
default: ;
}
break;
default: ;
}
if (!ChromaSubsampling.empty())
{
if (SamplingFactors.size()==4)
{
if (ChromaSubsampling=="4:4:4" && SamplingFactors[3].Hi==1 && SamplingFactors[3].Vi==1)
ChromaSubsampling+=":4";
else
ChromaSubsampling+=":?";
}
Fill(StreamKind_Last, 0, "ChromaSubsampling", ChromaSubsampling);
}
}
}
FILLING_END();
}
从代码的含义可知，提取出了图像的宽，高，采样方式等信息。
详细的代码暂时没有时间研究了，先这样了。