live555开发笔记（二）：live555创建RTSP服务器源码剖析，创建rtsp服务器的基本流程总结

前言

基于Live555的流媒体服务器方案源码剖析，了解基本的代码搭建步骤。

Demo

关于.h264与.265

  没深入研究，但是h264的后缀名.264替换为.h264文件，流媒体服务器连接播放rtsp是成功的。

阅读示例

这是学习的必经之路，live555本身是一套框架，我们需要按照预定义的流程建立服务，按照预定义的一些虚函数自己实现这些函数。文档和其他资料上这块都很少，所以自行参照示例进行逐步学习。

TestOnDemandRTSPServer.cpp示例研究

从main函数开始，遇到新的类则单独往下跳，分析完再回来，建议看至少2-3次，前后呼应分析过程。

全局变量

// 环境输出的
UsageEnvironment* env;
// 要使每个流的第二个和后续客户端重用与第一个客户端相同的输入流（而不是从每个客户端的开始播放文件），请将以下"False"更改为"True"
Boolean reuseFirstSource = False; // 要仅流式传输*MPEG-1 or 2视频"I"帧（例如，为了减少网络带宽），请将以下"False"更改为"True"： Boolean iFramesOnly = False; 

代码流程

首先创建任务调度器，然后设置我们的使用环境（注意：此代码是所有live555都需要这么做的）：

TaskScheduler* scheduler = BasicTaskScheduler::createNew(); env = BasicUsageEnvironment::createNew(*scheduler); 

开始进行权限控制，可以不控制则设置为null：

// 用于可选用户/密码身份验证的数据结构：
UserAuthenticationDatabase* authDB = NULL; 

控制是否可以权限控制的用户名和密码：

// 是否对接入权限进行控制
//#define ACCESS_CONTROL // 屏蔽了，则是没定义，不进行权限控制
#ifdef ACCESS_CONTROL
// 要实现对RTSP服务器的客户端访问控制，请执行以下操作：
authDB = new UserAuthenticationDatabase;
// 添加权限用于名称和密码，添加允许访问服务器的每个<username>、<password>
authDB->addUserRecord("username1", "password1");
// 对允许访问服务器的每个<username>、<password>重复上述操作。
#endif

创建RTSP服务：

// 创建RTSP服务，其中autoDB要是没有权限控制则是输入0(默认不写入就是输入0）
RTSPServer* rtspServer = RTSPServer::createNew(*env, 8554, authDB); if (rtspServer == NULL) { *env << "Failed to create RTSP server: " << env->getResultMsg() << "\n"; exit(1); } 

创建描述符，这个可以不用：

// 描述符，
char const* descriptionString = "Session streamed by \"testOnDemandRTSPServer\""; 

下面就是关键步骤了，添加RTSP服务器的流，整个流分为好几种，我们本章主要测试mp4文件，想挑mp4的来解说，但是没有mp4，那我们就直接挑一个.H264来解说：

// ===== 下面开始创建实际的流，有很多流，根据格式来 =====
// A MPEG-4 video elementary stream:
{
  char const* streamName = "mpeg4ESVideoTest"; char const* inputFileName = "test.m4e"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(MPEG4VideoFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A H.264 video elementary stream: { char const* streamName = "h264ESVideoTest"; char const* inputFileName = "test.264"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(H264VideoFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } 

其中announceStream函数只是打印流名称和文件名，不用管。其他的代码就是如何生成流，如何添加进去，后续大部分代码都是这样，至于具体的文件都是继承了几个主要类，笔者全部扫了一遍，本代码里面涉及的几个SubSession相关的子类，我们需要流媒体服务器处理数据，就是需要重写这些文件subsession来实现数据的处理，现在都截图出来，如下：

 
  
  
  
  

所以一共有三种：

• 继承自：FileServerMediaSubsession

• 继承自：OnDemandServerMediaSubsession

• 继承自：Medium

以上就是按照框架的方式设置一些参数，如帧率、分辨率、音频流数据、视频流数据，但是这个发的是视频流数据，编码还是解码都需要自己弄。

思考一下，如果是有一个本地文件，当作流媒体服务器，那么需要在有客户端获取rtsp流的时候，服务器就进行编码器准备开始解码，要数据的时候，就解码一帧数据往外发，大致就是这个流程。

这里对subsession不做深入探讨，继续分析接下来的代码：

// 尝试为RTSP over HTTP隧道创建一个HTTP服务器,首先尝试使用默认HTTP端口（80），然后使用其他HTTP端口号（8000和8080）
if(   rtspServer->setUpTunnelingOverHTTP(80) || rtspServer->setUpTunnelingOverHTTP(8000) || rtspServer->setUpTunnelingOverHTTP(8080)) { // 打印输出 *env << "\n(We use port " << rtspServer->httpServerPortNum() << " for optional RTSP-over-HTTP tunneling.)\n"; }else{ *env << "\n(RTSP-over-HTTP tunneling is not available.)\n"; } 

这是启动rtsp服务的http端口了。

// 启动流服务，不再返回了
env->taskScheduler().doEventLoop(); 

这是进入rtsp服务器循环了，程序将阻塞在此处了。

Live555流媒体服务实现基本流程

步骤一：创建RTSP服务器

创建一个Rtsp服务器实例，服务器实例后续需要添加流服务媒体实例。

步骤二：添加ServerMediaSession实例

创建服务媒体实例，这代表了一个流媒体对话，但是流媒体对话的流服务服务流需要单独创建，使用ServerMediaSubsession实例。

步骤三：实现ServerMediaSubSession实例（实现重点）

ServerMediaSubsession是具体服务流数的实例，在这个类中都需要实现流媒体的操作函数，这些操作函数决定了流媒体对应指令的具体参数配置和动作。

这里分为三类：

• 继承自：FileServerMediaSubsession
• 继承自：OnDemandServerMediaSubsession
• 继承自：Medium
    实现一些函数，咱们也不太清楚，但是知道需要拿到流的原始数据然后喂出去。
    

步骤四：启动服务端口

为rtsp创建一个http服务器，对外可以连接到的rtsp服务器。

步骤五：服务器运行

这是服务器进入循环，当前线程（单线程则是主线程，同时也可以代表本进程）则会阻塞在此处。

以上就是live555的rtsp建立rtsp服务器的流程，核心的点就在subsession中的数据处理，按照连接的流媒体格式，将流媒体的原始数据发出去，客户端就可以了播放了。

整理后的中文注释源码

#include "liveMedia.hh"
#include "BasicUsageEnvironment.hh" // webm mkv文件使用 static char newDemuxWatchVariable; static MatroskaFileServerDemux* matroskaDemux; static void onMatroskaDemuxCreation(MatroskaFileServerDemux* newDemux, void* /*clientData*/) { matroskaDemux = newDemux; newDemuxWatchVariable = 1; } // ogg文件使用 static OggFileServerDemux* oggDemux; static void onOggDemuxCreation(OggFileServerDemux* newDemux, void* /*clientData*/) { oggDemux = newDemux; newDemuxWatchVariable = 1; } // 打印流的相关信息 static void announceStream(RTSPServer* rtspServer, ServerMediaSession* sms, char const* streamName, char const* inputFileName) { // 获取rtsp地址 char* url = rtspServer->rtspURL(sms); // 获取使用环境 UsageEnvironment& env = rtspServer->envir(); // 使用环境输出"流名称"和"输入文件名" env << "\n\"" << streamName << "\" stream, from the file \"" << inputFileName << "\"\n"; env << "Play this stream using the URL \"" << url << "\"\n"; delete[] url; } int main(int argc, char** argv) { // 全局环境输出的 UsageEnvironment* env; // 要使每个流的第二个和后续客户端重用与第一个客户端相同的输入流（而不是从每个客户端的开始播放文件），请将以下"False"更改为"True" Boolean reuseFirstSource = False; // 要仅流式传输*MPEG-1 or 2视频"I"帧（例如，为了减少网络带宽），请将以下"False"更改为"True"： Boolean iFramesOnly = False; // 创建任务调度器、设置使用环境 TaskScheduler* scheduler = BasicTaskScheduler::createNew(); env = BasicUsageEnvironment::createNew(*scheduler); // 用于可选用户/密码身份验证的数据结构： UserAuthenticationDatabase* authDB = NULL; // 是否对接入权限进行控制 //#define ACCESS_CONTROL #ifdef ACCESS_CONTROL // 要实现对RTSP服务器的客户端访问控制，请执行以下操作： authDB = new UserAuthenticationDatabase; // 添加权限用于名称和密码，添加允许访问服务器的每个<username>、<password> authDB->addUserRecord("username1", "password1"); // 对允许访问服务器的每个<username>、<password>重复上述操作。 #endif // 创建RTSP服务，其中autoDB要是没有权限控制则是输入0(默认不写入就是输入0） RTSPServer* rtspServer = RTSPServer::createNew(*env, 8554, authDB); if(rtspServer == NULL) { *env << "Failed to create RTSP server: " << env->getResultMsg() << "\n"; exit(1); } // 描述符， char const* descriptionString = "Session streamed by \"testOnDemandRTSPServer\""; // 设置RTSP服务器可以提供的每个可能的流。 // 每个这样的流都是使用"ServerMediaSession"对象实现的， // 每个音频/视频子流都有一个或多个"ServerMediaSubsession"对象。 // ===== 下面开始创建实际的流，有很多流，根据格式来 ===== // A MPEG-4 video elementary stream: { char const* streamName = "mpeg4ESVideoTest"; char const* inputFileName = "test.m4e"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(MPEG4VideoFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A H.264 video elementary stream: { char const* streamName = "h264ESVideoTest"; char const* inputFileName = "test.264"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(H264VideoFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A H.265 video elementary stream: { char const* streamName = "h265ESVideoTest"; char const* inputFileName = "test.265"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(H265VideoFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A MPEG-1 or 2 audio+video program stream: { char const* streamName = "mpeg1or2AudioVideoTest"; char const* inputFileName = "test.mpg"; // NOTE: This *must* be a Program Stream; not an Elementary Stream ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); MPEG1or2FileServerDemux* demux = MPEG1or2FileServerDemux::createNew(*env, inputFileName, reuseFirstSource); sms->addSubsession(demux->newVideoServerMediaSubsession(iFramesOnly)); sms->addSubsession(demux->newAudioServerMediaSubsession()); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A MPEG-1 or 2 video elementary stream: { char const* streamName = "mpeg1or2ESVideoTest"; char const* inputFileName = "testv.mpg"; // NOTE: This *must* be a Video Elementary Stream; not a Program Stream ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(MPEG1or2VideoFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource, iFramesOnly)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A MP3 audio stream (actually, any MPEG-1 or 2 audio file will work): // To stream using 'ADUs' rather than raw MP3 frames, uncomment the following: //#define STREAM_USING_ADUS 1 // To also reorder ADUs before streaming, uncomment the following: //#define INTERLEAVE_ADUS 1 // (For more information about ADUs and interleaving, // see <http://www.live555.com/rtp-mp3/>) { char const* streamName = "mp3AudioTest"; char const* inputFileName = "test.mp3"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); Boolean useADUs = False; Interleaving* interleaving = NULL; #ifdef STREAM_USING_ADUS useADUs = True; #ifdef INTERLEAVE_ADUS unsigned char interleaveCycle[] = {0,2,1,3}; // or choose your own... unsigned const interleaveCycleSize = (sizeof interleaveCycle)/(sizeof (unsigned char)); interleaving = new Interleaving(interleaveCycleSize, interleaveCycle); #endif #endif sms->addSubsession(MP3AudioFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource, useADUs, interleaving)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A WAV audio stream: { char const* streamName = "wavAudioTest"; char const* inputFileName = "test.wav"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); // To convert 16-bit PCM data to 8-bit u-law, prior to streaming, // change the following to True: Boolean convertToULaw = False; sms->addSubsession(WAVAudioFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource, convertToULaw)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // An AMR audio stream: { char const* streamName = "amrAudioTest"; char const* inputFileName = "test.amr"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(AMRAudioFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A 'VOB' file (e.g., from an unencrypted DVD): { char const* streamName = "vobTest"; char const* inputFileName = "test.vob"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); // Note: VOB files are MPEG-2 Program Stream files, but using AC-3 audio MPEG1or2FileServerDemux* demux = MPEG1or2FileServerDemux::createNew(*env, inputFileName, reuseFirstSource); sms->addSubsession(demux->newVideoServerMediaSubsession(iFramesOnly)); sms->addSubsession(demux->newAC3AudioServerMediaSubsession()); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A MPEG-2 Transport Stream: { char const* streamName = "mpeg2TransportStreamTest"; char const* inputFileName = "test.ts"; char const* indexFileName = "test.tsx"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(MPEG2TransportFileServerMediaSubsession::createNew(*env, inputFileName, indexFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // An AAC audio stream (ADTS-format file): { char const* streamName = "aacAudioTest"; char const* inputFileName = "test.aac"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(ADTSAudioFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A DV video stream: { // First, make sure that the RTPSinks' buffers will be large enough to handle the huge size of DV frames (as big as 288000). OutPacketBuffer::maxSize = 300000; char const* streamName = "dvVideoTest"; char const* inputFileName = "test.dv"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(DVVideoFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A AC3 video elementary stream: { char const* streamName = "ac3AudioTest"; char const* inputFileName = "test.ac3"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(AC3AudioFileServerMediaSubsession::createNew(*env, inputFileName, reuseFirstSource)); rtspServer->addServerMediaSession(sms); announceStream(rtspServer, sms, streamName, inputFileName); } // A Matroska ('.mkv') file, with video+audio+subtitle streams: { char const* streamName = "matroskaFileTest"; char const* inputFileName = "test.mkv"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); newDemuxWatchVariable = 0; MatroskaFileServerDemux::createNew(*env, inputFileName, onMatroskaDemuxCreation, NULL); env->taskScheduler().doEventLoop(&newDemuxWatchVariable); Boolean sessionHasTracks = False; ServerMediaSubsession* smss; while ((smss = matroskaDemux->newServerMediaSubsession()) != NULL) { sms->addSubsession(smss); sessionHasTracks = True; } if (sessionHasTracks) { rtspServer->addServerMediaSession(sms); } // otherwise, because the stream has no tracks, we don't add a ServerMediaSession to the server. announceStream(rtspServer, sms, streamName, inputFileName); } // A WebM ('.webm') file, with video(VP8)+audio(Vorbis) streams: // (Note: ".webm' files are special types of Matroska files, so we use the same code as the Matroska ('.mkv') file code above.) { char const* streamName = "webmFileTest"; char const* inputFileName = "test.webm"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); newDemuxWatchVariable = 0; MatroskaFileServerDemux::createNew(*env, inputFileName, onMatroskaDemuxCreation, NULL); env->taskScheduler().doEventLoop(&newDemuxWatchVariable); Boolean sessionHasTracks = False; ServerMediaSubsession* smss; while((smss = matroskaDemux->newServerMediaSubsession()) != NULL) { sms->addSubsession(smss); sessionHasTracks = True; } if(sessionHasTracks) { rtspServer->addServerMediaSession(sms); } // otherwise, because the stream has no tracks, we don't add a ServerMediaSession to the server. announceStream(rtspServer, sms, streamName, inputFileName); } // An Ogg ('.ogg') file, with video and/or audio streams: { char const* streamName = "oggFileTest"; char const* inputFileName = "test.ogg"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); newDemuxWatchVariable = 0; OggFileServerDemux::createNew(*env, inputFileName, onOggDemuxCreation, NULL); env->taskScheduler().doEventLoop(&newDemuxWatchVariable); Boolean sessionHasTracks = False; ServerMediaSubsession* smss; while ((smss = oggDemux->newServerMediaSubsession()) != NULL) { sms->addSubsession(smss); sessionHasTracks = True; } if (sessionHasTracks) { rtspServer->addServerMediaSession(sms); } // otherwise, because the stream has no tracks, we don't add a ServerMediaSession to the server. announceStream(rtspServer, sms, streamName, inputFileName); } // An Opus ('.opus') audio file: // (Note: ".opus' files are special types of Ogg files, so we use the same code as the Ogg ('.ogg') file code above.) { char const* streamName = "opusFileTest"; char const* inputFileName = "test.opus"; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); newDemuxWatchVariable = 0; OggFileServerDemux::createNew(*env, inputFileName, onOggDemuxCreation, NULL); env->taskScheduler().doEventLoop(&newDemuxWatchVariable); Boolean sessionHasTracks = False; ServerMediaSubsession* smss; while((smss = oggDemux->newServerMediaSubsession()) != NULL) { sms->addSubsession(smss); sessionHasTracks = True; } if(sessionHasTracks) { rtspServer->addServerMediaSession(sms); } // otherwise, because the stream has no tracks, we don't add a ServerMediaSession to the server. announceStream(rtspServer, sms, streamName, inputFileName); } // A MPEG-2 Transport Stream, coming from a live UDP (raw-UDP or RTP/UDP) source: { char const* streamName = "mpeg2TransportStreamFromUDPSourceTest"; char const* inputAddressStr = "239.255.42.42"; // This causes the server to take its input from the stream sent by the "testMPEG2TransportStreamer" demo application. // (Note: If the input UDP source is unicast rather than multicast, then change this to NULL.) portNumBits const inputPortNum = 1234; // This causes the server to take its input from the stream sent by the "testMPEG2TransportStreamer" demo application. Boolean const inputStreamIsRawUDP = False; ServerMediaSession* sms = ServerMediaSession::createNew(*env, streamName, streamName, descriptionString); sms->addSubsession(MPEG2TransportUDPServerMediaSubsession::createNew(*env, inputAddressStr, inputPortNum, inputStreamIsRawUDP)); rtspServer->addServerMediaSession(sms); char* url = rtspServer->rtspURL(sms); *env << "\n\"" << streamName << "\" stream, from a UDP Transport Stream input source \n\t("; if (inputAddressStr != NULL) { *env << "IP multicast address " << inputAddressStr << ","; } else { *env << "unicast;"; } *env << " port " << inputPortNum << ")\n"; *env << "Play this stream using the URL \"" << url << "\"\n"; delete[] url; } // 尝试为RTSP over HTTP隧道创建一个HTTP服务器,首先尝试使用默认HTTP端口（80），然后使用其他HTTP端口号（8000和8080） if( rtspServer->setUpTunnelingOverHTTP(80) || rtspServer->setUpTunnelingOverHTTP(8000) || rtspServer->setUpTunnelingOverHTTP(8080)) { // 打印输出 *env << "\n(We use port " << rtspServer->httpServerPortNum() << " for optional RTSP-over-HTTP tunneling.)\n"; }else{ *env << "\n(RTSP-over-HTTP tunneling is not available.)\n"; } // 启动流服务，不再返回了 env->taskScheduler().doEventLoop(); return 0; } 

工程模板v1.1.0