关于对H264码流的TS的封装的相关代码实现

http://blog.csdn.net/max_min_go/article/details/39463675

1 写在开始之前

在前段时间有分享一个H264封装ps流到相关文章的，这次和大家分享下将H264封装成TS流到相关实现，其实也是工作工作需要。依照上篇一样，分段说明每个数据头的封装情况，当然，一样也会加上rtp头，方便以后的这方面到需求，如果开发不需要的话，可 以自行屏蔽掉，当然需要主要buffer指针的移动情况

2 封装的各个头到规则要点

整个封装过程也是和ps类似,但是最大到区别在于TS流到数据长度都是固定188大小来传输的，而PS流则是可变包结构，正因为两者在结构上到差异，导致了它们在传输误码上的不同抵抗力.TS流由于采用了固定长度的数据包，当传输误码破坏了某一个TS包的同步信息时，接收端可在固定的位置检测到下一个TS包到同步信息，从而恢复同步，避免数据丢失,PS流则长度可变到数据包，当某一个ps包同步信息丢失时，接收端就无法进行信息同步，也就无法确认下一步到同步信息，导致了严重到信息丢失。因此在环境恶劣的情况下， 传输码丢失比较严重，一般都采用TS流来进行避免，当网络环境比较稳定，传输误码概率小，这个时候采用PS流进行传送。

关于TS流需要了解下节目映射表(PAT：Program Associate Table)以及节目映射表(PMT：Program Map Table)，当发送到数据为视频数据关键帧的时候，需要在包头中添加PAT和PMT

具体结构体如下

封装组成：(PAT +PMT) + TS + PES + H264 + (TS + H264 + TS + H264 ....)

数据长度：PES包的长度=188字节－TS包头长度（4字节）－适应域长度(PES长度或者0)

注意：
a 每次数据定长包188个字节，如果不足的则用1填充，这里填充时值每一个bit位都填充，memset就是最好选择。

b 因为我个人习惯，在封装到时候当为关键帧的时候，我直接丢了PAM+PMT+TS+PES 然后填充满188个字节，这样做提醒大家 是错误到，完全错误的，PES后必须跟H264数据。

c PES能表示的数据长度只有short, 两个字节，所以当数据长度超过的话，则需要考虑多个PES头

3 各个部件到头的伪代码实现

[cpp] view
plain copy

/*

*@remark: 整体发送数据的抽象逻辑处理函数接口

*/

int rtsp_RTPPackage( RTP_SESSION_S *pRtpSender, int nFrameLen, StreamType_E enStreamType)

{

int nRet = 0;

int bVideo = 1 ;

int nSendDataOff = 0;

int nSendSize = 0;

int nPayLoadSize = 0;

int nHasSend = 0;

int IFrameFlag = 0;

char TSFrameHdr[1024];

int nHead = 0;

memset(TSFrameHdr, 0, 1024);

memset(pRtpSender->stRtpPack, 0, RTP_MAX_PACKET_BUFF);

bVideo = ((enStreamType == VIDEO_STREAM ) ? 1 : 0);

// @remark: 判断数据是否为I帧，如果为I帧的话，加上PAT和PMT

if( (bVideo == 1) && pRtpSender->stAvData.u8IFrame == 1)

{

if((nRet = mk_ts_pat_packet(TSFrameHdr +nSendDataOff,

pRtpSender->hHdlTs)) <= 0)

{

DBG_INFO(" mk_ts_pat_packet failed!\n");

return -1;

}

// @remark: 每次添加一个头的时候，必须注意指针到偏移量

nSendDataOff += nRet;

if((nRet = mk_ts_pmt_packet(TSFrameHdr + nSendDataOff,

pRtpSender->hHdlTs)) <= 0)

{

DBG_INFO(" mk_ts_pmt_packet failed!\n");

return -1;

}

nSendDataOff += nRet;

}

// @remark: 添加PS头，需要注意ps里也有一个计数的字段

if((nRet = mk_ts_packet(TSFrameHdr + nSendDataOff, pRtpSender->hHdlTs,

1, bVideo, pRtpSender->stAvData.u8IFrame, pRtpSender->stAvData.u64TimeStamp)) <= 0 )

{

DBG_INFO(" mk_ts_packet failed!\n");

return -1;

}

nSendDataOff += nRet;

//此字段是用来计算ts长度，因为ts包是固定188字节长度

nHead = nRet;

// @remark: 添加PES头，后面就必须接H264数据了，不能通过1来填充

if((nRet = mk_pes_packet(TSFrameHdr + nSendDataOff, bVideo, nFrameLen, 1,

pRtpSender->stAvData.u64TimeStamp, pRtpSender->stAvData.u64TimeStamp)) <= 0 )

{

DBG_INFO(" mk_pes_packet failed!\n");

return -1;

}

nSendDataOff += nRet;

nHead += nRet;

// @remark: 如果第一次发送的数据长度大于剩余长度，则先发送ts包剩余长度的数据

if( nFrameLen > (TS_LOAD_LEN - nHead))

{

memcpy(TSFrameHdr + nSendDataOff, pRtpSender->stAvData.data, TS_LOAD_LEN - nHead);

nSendDataOff += (TS_LOAD_LEN - nHead);

nHasSend = (TS_LOAD_LEN - nHead);

if( rtsp_send_rtppack(TSFrameHdr, &nSendDataOff, pRtpSender->stAvData.u64TimeStamp, 0, (pRtpSender->stAvData.u8IFrame?1:0), bVideo, 1, pRtpSender) != 0 )

{

DBG_INFO(" rtsp_send_pack failed!\n");

return -1;

}

}

// @remark: 如果第一次发送数据长度小于ts头剩余长度，则，发送数据帧长度，剩余没有188长度的用1填充

else

{

memcpy(TSFrameHdr + nSendDataOff, pRtpSender->stAvData.data, nFrameLen);

nSendDataOff += nFrameLen;

nHasSend = nFrameLen;

memset(TSFrameHdr +nSendDataOff, 0xFF, (TS_LOAD_LEN-nHead - nFrameLen));

nSendDataOff += (TS_LOAD_LEN -nHead- nFrameLen);

if( rtsp_send_rtppack(TSFrameHdr, &nSendDataOff, pRtpSender->stAvData.u64TimeStamp, 1, (pRtpSender->stAvData.u8IFrame?1:0), bVideo, 1, pRtpSender) != 0 )

{

DBG_INFO(" rtsp_send_rtppack failed!\n");

return -1;

}

}

// 对应的数据便宜长度，因为我处理的时候时固定1460到rtp包发送数据，所以这里会处理偏移，方便添加rtp头

nPayLoadSize = RTP_MAX_PACKET_BUFF - 4 - RTP_HDR_LEN - (4+6) * 7; // 减去rtp头，ts头 ,一个rtp包最多7个ts包

nFrameLen -= (TS_LOAD_LEN - nHead);

// @remark: 第二次发送数据了，此时发送数据时候，就需要外再添加ps头了

while(nFrameLen > 0 )

{

nSendSize = (nFrameLen > nPayLoadSize) ? nPayLoadSize : nFrameLen;

if( rtsp_send_rtppack(pRtpSender->stAvData.data + nHasSend, &nSendSize, pRtpSender->stAvData.u64TimeStamp,

((nSendSize == nFrameLen) ? 1 : 0), IFrameFlag, bVideo, 0, pRtpSender) != 0 )

{

DBG_INFO(" rtsp_send_rtppack failed!\n");

return -1;

}

nFrameLen -= nSendSize;

nHasSend += nSendSize;

memset(pRtpSender->stRtpPack, 0, RTP_MAX_PACKET_BUFF);

IFrameFlag = 0;

}

return 0;

}

[cpp] view
plain copy

/*

*@remark : 添加pat头

*/

int mk_ts_pat_packet(char *buf, int handle)

{

int nOffset = 0;

int nRet = 0;

if (!buf)

{

return 0;

}

if (0 >= (nRet = ts_header(buf, handle, TS_TYPE_PAT, 1)))

{

return 0;

}

nOffset += nRet;

if (0 >= (nRet = ts_pointer_field(buf + nOffset)))

{

return 0;

}

nOffset += nRet;

if (0 >= (nRet = ts_pat_header(buf + nOffset)))

{

return 0;

}

nOffset += nRet;

// 每一个pat都会当成一个ts包来处理，所以每次剩余部分用1来充填完

memset(buf + nOffset, 0xFF, TS_PACKET_SIZE - nOffset);

return TS_PACKET_SIZE;

}

int ts_pat_header(char *buf)

{

BITS_BUFFER_S bits;

if (!buf)

{

return 0;

}

bits_initwrite(&bits, 32, (unsigned char *)buf);

bits_write(&bits, 8, 0x00); // table id, 固定为0x00

bits_write(&bits, 1, 1); // section syntax indicator, 固定为1

bits_write(&bits, 1, 0); // zero, 0

bits_write(&bits, 2, 0x03); // reserved1, 固定为0x03

bits_write(&bits, 12, 0x0D); // section length, 表示这个字节后面有用的字节数, 包括CRC32

bits_write(&bits, 16, 0x0001); // transport stream id, 用来区别其他的TS流

bits_write(&bits, 2, 0x03); // reserved2, 固定为0x03

bits_write(&bits, 5, 0x00); // version number, 范围0-31

bits_write(&bits, 1, 1); // current next indicator, 0 下一个表有效, 1当前传送的PAT表可以使用

bits_write(&bits, 8, 0x00); // section number, PAT可能分为多段传输，第一段为00

bits_write(&bits, 8, 0x00); // last section number

bits_write(&bits, 16, 0x0001); // program number

bits_write(&bits, 3, 0x07); // reserved3和pmt_pid是一组，共有几个频道由program number指示

bits_write(&bits, 13, TS_PID_PMT); // pmt of pid in ts head

bits_write(&bits, 8, 0x9F); // CRC_32 先暂时写死

bits_write(&bits, 8, 0xC7);

bits_write(&bits, 8, 0x62);

bits_write(&bits, 8, 0x58);

bits_align(&bits);

return bits.i_data;

}

[cpp] view
plain copy

/*

*@remaark: 添加PMT头

*/

int mk_ts_pmt_packet(char *buf, int handle)

{

int nOffset = 0;

int nRet = 0;

if (!buf)

{

return 0;

}

if (0 >= (nRet = ts_header(buf, handle, TS_TYPE_PMT, 1)))

{

return 0;

}

nOffset += nRet;

if (0 >= (nRet = ts_pointer_field(buf + nOffset)))

{

return 0;

}

nOffset += nRet;

if (0 >= (nRet = ts_pmt_header(buf + nOffset)))

{

return 0;

}

nOffset += nRet;

// 每一个pmt都会当成一个ts包来处理，所以每次剩余部分用1来充填完

memset(buf + nOffset, 0xFF, TS_PACKET_SIZE - nOffset);

return TS_PACKET_SIZE;

}

int ts_pmt_header(char *buf)

{

BITS_BUFFER_S bits;

if (!buf)

{

return 0;

}

bits_initwrite(&bits, 32, (unsigned char *)buf);

bits_write(&bits, 8, 0x02); // table id, 固定为0x02

bits_write(&bits, 1, 1); // section syntax indicator, 固定为1

bits_write(&bits, 1, 0); // zero, 0

bits_write(&bits, 2, 0x03); // reserved1, 固定为0x03

bits_write(&bits, 12, 0x1C); // section length, 表示这个字节后面有用的字节数, 包括CRC32

bits_write(&bits, 16, 0x0001); // program number, 表示当前的PMT关联到的频道号码

bits_write(&bits, 2, 0x03); // reserved2, 固定为0x03

bits_write(&bits, 5, 0x00); // version number, 范围0-31

bits_write(&bits, 1, 1); // current next indicator, 0 下一个表有效, 1当前传送的PAT表可以使用

bits_write(&bits, 8, 0x00); // section number, PAT可能分为多段传输，第一段为00

bits_write(&bits, 8, 0x00); // last section number

bits_write(&bits, 3, 0x07); // reserved3, 固定为0x07

bits_write(&bits, 13, TS_PID_VIDEO); // pcr of pid in ts head, 如果对于私有数据流的节目定义与PCR无关，这个域的值将为0x1FFF

bits_write(&bits, 4, 0x0F); // reserved4, 固定为0x0F

bits_write(&bits, 12, 0x00); // program info length, 前两位bit为00

bits_write(&bits, 8, TS_PMT_STREAMTYPE_H264_VIDEO); // stream type, 标志是Video还是Audio还是其他数据

bits_write(&bits, 3, 0x07); // reserved, 固定为0x07

bits_write(&bits, 13, TS_PID_VIDEO); // elementary of pid in ts head

bits_write(&bits, 4, 0x0F); // reserved, 固定为0x0F

bits_write(&bits, 12, 0x00); // elementary stream info length, 前两位bit为00

bits_write(&bits, 8, TS_PMT_STREAMTYPE_11172_AUDIO); // stream type, 标志是Video还是Audio还是其他数据

bits_write(&bits, 3, 0x07); // reserved, 固定为0x07

bits_write(&bits, 13, TS_PID_AUDIO); // elementary of pid in ts head

bits_write(&bits, 4, 0x0F); // reserved, 固定为0x0F

bits_write(&bits, 12, 0x00); // elementary stream info length, 前两位bit为00

bits_write(&bits, 8, 0xA4); // stream type, 标志是Video还是Audio还是其他数据

bits_write(&bits, 3, 0x07); // reserved, 固定为0x07

bits_write(&bits, 13, 0x00A4); // elementary of pid in ts head

bits_write(&bits, 4, 0x0F); // reserved, 固定为0x0F

bits_write(&bits, 12, 0x00); // elementary stream info length, 前两位bit为00

bits_write(&bits, 8, 0x34); //CRC_32 先暂时写死

bits_write(&bits, 8, 0x12);

bits_write(&bits, 8, 0xA3);

bits_write(&bits, 8, 0x72);

bits_align(&bits);

return bits.i_data;

}

[cpp] view
plain copy

/*

*@remark: ts头的封装

*/

int mk_ts_packet(char *buf, int handle, int bStart, int bVideo, int bIFrame, unsigned long long timestamp)

{

int nOffset = 0;

int nRet = 0;

if (!buf)

{

return 0;

}

if (0 >= (nRet = ts_header(buf, handle, bVideo ? TS_TYPE_VIDEO : TS_TYPE_AUDIO, bStart)))

{

return 0;

}

nOffset += nRet;

if (0 >= (nRet = ts_adaptation_field(buf + nOffset, bStart, bVideo && (bIFrame), timestamp)))

{

return 0;

}

nOffset += nRet;

return nOffset;

}

/* *@remark: ts头相关封装

* PSI 包括了PAT、PMT、NIT、CAT

* PSI--Program Specific Information, PAT--program association table, PMT--program map table

* NIT--network information table, CAT--Conditional Access Table

* 一个网络中可以有多个TS流(用PAT中的ts_id区分)

* 一个TS流中可以有多个频道(用PAT中的pnumber、pmt_pid区分)

* 一个频道中可以有多个PES流(用PMT中的mpt_stream区分)

*/

int ts_header(char *buf, int handle, TS_TYPE_E type, int bStart)

{

BITS_BUFFER_S bits;

TS_MNG_S *pMng = (TS_MNG_S *)handle;

if (!buf || !handle || TS_TYPE_BEGIN >= type || TS_TYPE_END <= type)

{

return 0;

}

bits_initwrite(&bits, 32, (unsigned char *)buf);

bits_write(&bits, 8, 0x47); // sync_byte, 固定为0x47,表示后面的是一个TS分组

// payload unit start indicator根据TS packet究竟是包含PES packet还是包含PSI data而设置不同值

// 1. 若包含的是PES packet header, 设为1, 如果是PES packet余下内容, 则设为0

// 2. 若包含的是PSI data, 设为1, 则payload的第一个byte将是point_field, 0则表示payload中没有point_field

// 3. 若此TS packet为null packet, 此flag设为0

bits_write(&bits, 1, 0); // transport error indicator

bits_write(&bits, 1, bStart); // payload unit start indicator

bits_write(&bits, 1, 0); // transport priority, 1表示高优先级

if (TS_TYPE_PAT == type)

{

bits_write(&bits, 13, 0x00); // pid, 0x00 PAT, 0x01 CAT

}

else if (TS_TYPE_PMT == type)

{

bits_write(&bits, 13, TS_PID_PMT);

}

else if (TS_TYPE_VIDEO == type)

{

bits_write(&bits, 13, TS_PID_VIDEO);

}

else if (TS_TYPE_AUDIO == type)

{

bits_write(&bits, 13, TS_PID_AUDIO);

}

bits_write(&bits, 2, 0); // transport scrambling control, 传输加扰控制

if (TS_TYPE_PAT == type || TS_TYPE_PMT == type)

{

// continuity counter, 是具有同一PID值的TS包之间的连续计数值

// 当分组的adaption_field_control字段为00话10时，该字段不递增

bits_write(&bits, 2, 0x01); // adaptation field control, 00 forbid, 01 have payload, 10 have adaptation, 11 have payload and adaptation

bits_write(&bits, 4, pMng->nPatCounter); // continuity counter, 0~15

if (TS_TYPE_PAT != type)

{

pMng->nPatCounter++;

pMng->nPatCounter &= 0x0F;

}

}

else

{

bits_write(&bits, 2, 0x03); // 第一位表示有无调整字段，第二位表示有无有效负载

bits_write(&bits, 4, pMng->nContinuityCounter);

pMng->nContinuityCounter++;

pMng->nContinuityCounter &= 0x0F;

}

bits_align(&bits);

return bits.i_data;

}

[cpp] view
plain copy

/*

*remark:添加pes头

*/

int mk_pes_packet(char *buf, int bVideo, int length, int bDtsEn, unsigned long long pts, unsigned long long dts)

{

PES_HEAD_S pesHead;

PES_OPTION_S pesOption;

PES_PTS_S pesPts;

PES_PTS_S pesDts;

if (!buf)

{

return 0;

}

if( bVideo == 1)

{

// 视频的采样频率为90kHZ,则增量为3600

pts = pts * 9 / 100; // 90000Hz

dts = dts * 9 / 100; // 90000Hz

}

else

{

// 音频的话，则需要按照8000HZ来计算增量[需要的话]

pts = pts * 8 / 1000; // 8000Hz

dts = dts * 8 / 1000; // 8000Hz

}

memset(&pesHead, 0, sizeof(pesHead));

memset(&pesOption, 0, sizeof(pesOption));

memset(&pesPts, 0, sizeof(pesPts));

memset(&pesDts, 0, sizeof(pesDts));

pesHead.startcode = htonl(0x000001) >> 8;

pesHead.stream_id = bVideo ? 0xE0 : 0xC0;

if (PES_MAX_SIZE < length)

{

pesHead.pack_len = 0;

}

else

{

pesHead.pack_len = htons(length + sizeof(pesOption) + sizeof(pesPts) + (bDtsEn ? sizeof(pesDts) : 0));

}

pesOption.fixed = 0x02;

pesOption.pts_dts = bDtsEn ? 0x03 : 0x02;

pesOption.head_len = sizeof(pesPts) + (bDtsEn ? sizeof(pesDts) : 0);

pesPts.fixed2 = pesPts.fixed3 = pesPts.fixed4 = 0x01;

pesPts.fixed1 = bDtsEn ? 0x03 : 0x02;

pesPts.ts1 = (pts >> 30) & 0x07;

pesPts.ts2 = (pts >> 22) & 0xFF;

pesPts.ts3 = (pts >> 15) & 0x7F;

pesPts.ts4 = (pts >> 7) & 0xFF;

pesPts.ts5 = pts & 0x7F;

pesDts.fixed1 = pesDts.fixed2 = pesDts.fixed3 = pesDts.fixed4 = 0x01;

pesDts.ts1 = (dts >> 30) & 0x07;

pesDts.ts2 = (dts >> 22) & 0xFF;

pesDts.ts3 = (dts >> 15) & 0x7F;

pesDts.ts4 = (dts >> 7) & 0xFF;

pesDts.ts5 = dts & 0x7F;

char *head = buf;

memcpy(head, &pesHead, sizeof(pesHead));

head += sizeof(pesHead);

memcpy(head, &pesOption, sizeof(pesOption));

head += sizeof(pesOption);

memcpy(head, &pesPts, sizeof(pesPts));

head += sizeof(pesPts);

if (bDtsEn)

{

memcpy(head, &pesDts, sizeof(pesDts));

head += sizeof(pesPts);

}

return (head - buf);

}

[cpp] view
plain copy

/*

*@remark: 最后封装rtp头并发送最终封装好到完整的数据包

*/

int rtsp_send_rtppack(char *Databuf, int *datalen, unsigned long curtimestamp, int mark_flag, int IFrameFlag, int bVideo, int nFrameStart, RTP_SESSION_S *pRtpSender)

{

int nHasSend = 0;

int nRet = 0;

int nTsHeadNum = 0;

int nHadDataLen = 0;

int nTcpSendLen = 0;

static unsigned short cSeqnum;

// @remark：表示为数据的第一次发送，所以不需要额外再添加ts头

if( nFrameStart == 1 )

{

nRet = mk_rtp_packet(pRtpSender->stRtpPack + nHasSend, mark_flag, IFrameFlag, bVideo, ++cSeqnum, (curtimestamp * 9/100));

nHasSend += nRet;

memcpy(pRtpSender->stRtpPack + nHasSend, Databuf, *datalen);

nHasSend += *datalen;

}

else // 不是第一次发送此帧数据的话，则需要添加封装新的ts包，并添加ts头

{

// rtp+ rtp_ext + ts +data

nRet = mk_rtp_packet(pRtpSender->stRtpPack + nHasSend, mark_flag, IFrameFlag, bVideo, ++cSeqnum, (curtimestamp * 9/100));

nHasSend += nRet;

while(*datalen > 0 && nTsHeadNum < 7)

{

nRet = mk_ts_packet(pRtpSender->stRtpPack + nHasSend , pRtpSender->hHdlTs, 0, bVideo, (IFrameFlag > 0 ? 1:0), curtimestamp);

nHasSend += nRet;

if(*datalen < (TS_LOAD_LEN- nRet))

{

memcpy(pRtpSender->stRtpPack + nHasSend, Databuf + nHadDataLen, *datalen);

nHasSend += *datalen;

nHadDataLen += *datalen;

//不够Ts188用1补充

memset(pRtpSender->stRtpPack + nHasSend, 0xFF, TS_LOAD_LEN- nRet - (*datalen));

nHasSend += (TS_LOAD_LEN - nRet - *datalen);

}

else

{

memcpy(pRtpSender->stRtpPack + nHasSend, Databuf + nHadDataLen, TS_LOAD_LEN - nRet);

nHasSend += (TS_LOAD_LEN - nRet);

*datalen -= (TS_LOAD_LEN - nRet);

nHadDataLen += (TS_LOAD_LEN - nRet);

}

nTsHeadNum ++;

}

*datalen = nHadDataLen; //实际发送裸数据到长度

}

if(pRtpSender->RtspsockFd <= 0 )

{

DBG_INFO("send rtp packet socket error\n");

return -1;

}

nTcpSendLen = hi_tcp_noblock_send(pRtpSender->RtspsockFd, pRtpSender->stRtpPack, nHasSend, NULL,1500);

if(nTcpSendLen != nHasSend )

{

DBG_INFO("send rtp packet failed:%s\n",strerror(errno));

return -1;

}

return 0;

}

[cpp] view
plain copy

/*

*remark: 上面用到的一些宏定义和一些关于字节操作的函数，很多一些开源到视频处理的库都能看到，

为了方便也都将贴出来分享，当然也可以参考下vlc里面的源码

*/

/*@remark: 常量定义 */

#define TS_PID_PMT (0x62)

#define TS_PID_VIDEO (0x65)

#define TS_PID_AUDIO (0x84)

#define TS_PMT_STREAMTYPE_11172_AUDIO (0x03)

#define TS_PMT_STREAMTYPE_13818_AUDIO (0x04)

#define TS_PMT_STREAMTYPE_AAC_AUDIO (0x0F)

#define TS_PMT_STREAMTYPE_H264_VIDEO (0x1B)

/* @remark: 结构体定义 */

typedef struct

{

int i_size; // p_data字节数

int i_data; // 当前操作字节的位置

unsigned char i_mask; // 当前操作位的掩码

unsigned char *p_data; // bits buffer

} BITS_BUFFER_S;

typedef struct

{

unsigned int startcode : 24; // 固定为00 00 01

unsigned int stream_id : 8; // 0xC0-0xDF audio stream, 0xE0-0xEF video stream, 0xBD Private stream 1, 0xBE Padding stream, 0xBF Private stream 2

unsigned short pack_len; // PES packet length

} __attribute__ ((packed)) PES_HEAD_S;

typedef struct

{

#if (BYTE_ORDER == LITTLE_ENDIAN)

unsigned char original : 1; // original or copy, 原版或拷贝

unsigned char copyright : 1; // copyright flag

unsigned char align : 1; // data alignment indicator, 数据定位指示符

unsigned char priority : 1; // PES priority

unsigned char scramb : 2; // PES Scrambling control, 加扰控制

unsigned char fixed : 2; // 固定为10

unsigned char exten : 1; // PES extension flag

unsigned char crc : 1; // PES CRC flag

unsigned char acopy : 1; // additional copy info flag

unsigned char trick : 1; // DSM(Digital Storage Media) trick mode flag

unsigned char rate : 1; // ES rate flag, ES流速率标志

unsigned char escr : 1; // ESCR(Elementary Stream Clock Reference) flag, ES流时钟基准标志

unsigned char pts_dts : 2; // PTS DTS flags, 00 no PTS and DTS, 01 forbid, 10 have PTS, 11 have PTS and DTS

#elif (BYTE_ORDER == BIG_ENDIAN)

unsigned char fixed : 2; // 固定为10

unsigned char scramb : 2; // PES Scrambling control, 加扰控制

unsigned char priority : 1; // PES priority

unsigned char align : 1; // data alignment indicator, 数据定位指示符

unsigned char copyright : 1; // copyright flag

unsigned char original : 1; // original or copy, 原版或拷贝

unsigned char pts_dts : 2; // PTS DTS flags, 00 no PTS and DTS, 01 forbid, 10 have PTS, 11 have PTS and DTS

unsigned char escr : 1; // ESCR(Elementary Stream Clock Reference) flag, ES流时钟基准标志

unsigned char rate : 1; // ES rate flag, ES流速率标志

unsigned char trick : 1; // DSM(Digital Storage Media) trick mode flag

unsigned char acopy : 1; // additional copy info flag

unsigned char crc : 1; // PES CRC flag

unsigned char exten : 1; // PES extension flag

#endif

unsigned char head_len; // PES header data length

} __attribute__ ((packed)) PES_OPTION_S;

typedef struct

{// ts total 33 bits

#if (BYTE_ORDER == LITTLE_ENDIAN)

unsigned char fixed2 : 1; // 固定为1

unsigned char ts1 : 3; // bit30-32

unsigned char fixed1 : 4; // DTS为0x01, PTS为0x02, PTS+DTS则PTS为0x03

unsigned char ts2; // bit22-29

unsigned char fixed3 : 1; // 固定为1

unsigned char ts3 : 7; // bit15-21

unsigned char ts4; // bit7-14

unsigned char fixed4 : 1; // 固定为1

unsigned char ts5 : 7; // bit0-6

#elif (BYTE_ORDER == BIG_ENDIAN)

unsigned char fixed1 : 4; // DTS为0x01, PTS为0x02, PTS+DTS则PTS为0x03

unsigned char ts1 : 3; // bit30-32

unsigned char fixed2 : 1; // 固定为1

unsigned char ts2; // bit22-29

unsigned char ts3 : 7; // bit15-21

unsigned char fixed3 : 1; // 固定为1

unsigned char ts4; // bit7-14

unsigned char ts5 : 7; // bit0-6

unsigned char fixed4 : 1; // 固定为1

#endif

} __attribute__ ((packed)) PES_PTS_S;

/* remark:接口函数定义 */

int bits_initwrite(BITS_BUFFER_S *p_buffer, int i_size, unsigned char *p_data)

{

if (!p_data)

{

return -1;

}

p_buffer->i_size = i_size;

p_buffer->i_data = 0;

p_buffer->i_mask = 0x80;

p_buffer->p_data = p_data;

p_buffer->p_data[0] = 0;

return 0;

}

void bits_align(BITS_BUFFER_S *p_buffer)

{

if (p_buffer->i_mask != 0x80 && p_buffer->i_data < p_buffer->i_size)

{

p_buffer->i_mask = 0x80;

p_buffer->i_data++;

p_buffer->p_data[p_buffer->i_data] = 0x00;

}

}

inline void bits_write(BITS_BUFFER_S *p_buffer, int i_count, unsigned long i_bits)

{

while (i_count > 0)

{

i_count--;

if ((i_bits >> i_count ) & 0x01)

{

p_buffer->p_data[p_buffer->i_data] |= p_buffer->i_mask;

}

else

{

p_buffer->p_data[p_buffer->i_data] &= ~p_buffer->i_mask;

}

p_buffer->i_mask >>= 1;

if (p_buffer->i_mask == 0)

{

p_buffer->i_data++;

p_buffer->i_mask = 0x80;

}

}

}

int bits_initread(BITS_BUFFER_S *p_buffer, int i_size, unsigned char *p_data)

{

if (!p_data)

{

return -1;

}

p_buffer->i_size = i_size;

p_buffer->i_data = 0;

p_buffer->i_mask = 0x80;

p_buffer->p_data = p_data;

return 0;

}

inline int bits_read(BITS_BUFFER_S *p_buffer, int i_count, unsigned long *i_bits)

{

if (!i_bits)

{

return -1;

}

*i_bits = 0;

while (i_count > 0)

{

i_count--;

if (p_buffer->p_data[p_buffer->i_data] & p_buffer->i_mask)

{

*i_bits |= 0x01;

}

if (i_count)

{

*i_bits = *i_bits << 1;

}

p_buffer->i_mask >>= 1;

if(p_buffer->i_mask == 0)

{

p_buffer->i_data++;

p_buffer->i_mask = 0x80;

}

}

return 0;

}

5 写在最后

看过我上一篇的关于ps封装的可能会注意的，关于压字节的处理，两篇博文到处理方式有些差异。关于我这个我简单说两点

第一次是这个ts的处理里面封装是另外一个同事实现的，我因为用到所以拿来使用，但是上次调用封装都是自己完成。第二个就是

ps和ts的处理方式不一样。一个定长，一个不定长。所以这样处理，也挺好的，我也有点懒，所以没有改过来。