【RTSP从零实践】12、TCP传输H264格式RTP包(RTP_over_TCP)的RTSP服务器(附带源码)
😁博客主页😁:🚀https://blog.csdn.net/wkd_007🚀
🤑博客内容🤑:🍭嵌入式开发、Linux、C语言、C++、数据结构、音视频🍭
🤣本文内容🤣:🍭TCP传输H264格式RTP包(RTP_OVER_TCP)的RTSP服务器🍭
⏰发布时间⏰: 2025-07-16
本文未经允许,不得转发!!!
目录
- 🎄一、概述
- 🎄二、RTP over TCP(RTSP) 介绍
- ✨2.1 RTP over TCP(RTSP) 相关概念
- ✨2.2 怎么区分 RTSP包、RTP包、RTCP包
- ✨2.3 RTP包封装代码
- 🎄三、实现步骤、实现细节
- ✨3.1 创建TCP服务端套接字
- ✨3.2 处理RTSP命令
- ✨3.3 读取H264文件
- ✨3.4 封装成RTP包并发送
- 🎈3.4.1、RTP包最开始的4个字节
- 🎈3.4.2、RTP头(RTP Header)
- 🎈3.4.3、H264 的 RTP负载(RTP Payload)
- 🎄四、RTP_over_TCP的RTSP服务端源码
- 🎄五、、总结
前面系列文章回顾:
【音视频 | RTSP】RTSP协议详解 及 抓包例子解析(详细而不赘述)
【音视频 | RTSP】SDP(会话描述协议)详解 及 抓包例子分析
【音视频 | RTP】RTP协议详解(H.264的RTP封包格式、AAC的RTP封包格式)
【RTSP从零实践】01、根据RTSP协议实现一个RTSP服务
【RTSP从零实践】02、使用RTP协议封装并传输H264
【RTSP从零实践】03、实现最简单的传输H264的RTSP服务器
【RTSP从零实践】04、使用RTP协议封装并传输AAC
【RTSP从零实践】05、实现最简单的传输AAC的RTSP服务器
【RTSP从零实践】06、实现最简单的同时传输H264、AAC的RTSP服务器
【RTSP从零实践】07、多播传输H264格式的RTP包(附带源码)
【RTSP从零实践】08、多播传输H264码流的RTSP服务器——最简单的实现例子(附带源码)
【RTSP从零实践】09、多播传输AAC格式的RTP包(附带源码)
【RTSP从零实践】10、多播传输AAC码流的RTSP服务器——最简单的实现例子(附带源码)
【RTSP从零实践】11、多播同时传输H264、AAC码流的RTSP服务器——最简单的实现例子(附带源码)
🎄一、概述
前面文章介绍的文章都是由UDP作为传输层协议来发送RTP报文的(RTP over UDP
),在RTSP服务器中,还有一种RTP包传输方式,即使用TCP协议作为传输层协议发送RTP报文的RTP over TCP
。本文将介绍怎样实现一个最简单的使用TCP方式发送RTP包的RTSP服务器。
本文内容安排如下:
- 1、介绍使用tcp协议发送RTP(
RTP over TCP
)的相关内容: - 2、使用tcp协议发送RTP的RTSP服务端实现步骤和细节。
🎄二、RTP over TCP(RTSP) 介绍
✨2.1 RTP over TCP(RTSP) 相关概念
我们前面系列文章介绍过的rtsp服务器都是创建了一个TCP服务来处理RTSP协议,创建另一个UDP套接字来发送RTP包,这种方式就是 RTP over UDP
。而RTP over TCP
不单独建立一个UDP套接字去发送RTP包,而是利用处理RTSP协议的TCP套接字来发送RTP包的,所有有些资料也把这种方式称为RTP over RTSP
。
RTP over UDP
:一个TCP套接字处理RTSP协议,另一个UDP套接字发送RTP包、RTCP包;RTP over TCP(RTSP)
:一个TCP套接字处理RTSP协议、RTP包、RTCP包。
✨2.2 怎么区分 RTSP包、RTP包、RTCP包
如上面所说,我们复用发送RTSP交互的socket来发送RTP包和RTCP信息,那么对于客户端来说,如何区分这三种数据呢?
我们将这三个分为两类,一类是RTSP,一类是RTP、RTCP
发送RTSP信息的情况没有变化,还是更以前一样的方式
发送RTP、RTCP包,在每个包前面都加上四个字节,这四个字节解释如下:
字节 | 描述 |
---|---|
第一个字节 | 字符'$' ,表示这个包是RTP包 或 RTCP包 |
第二个字节 | 通道号channel,用于区分RTP包 或 RTCP包 |
第三、四个字节 | 表示RTP包的大小 |
使用TCP协议传输的RTP包和RTCP包,第一个字节固定为$
字符,第二字节的channel是在RTSP服务器处理的SETUP过程中,客户端发送给服务端的。
所以,使用TCP协议传输的RTP包结构如下:
✨2.3 RTP包封装代码
在RTP包起始位置增加4个字节的数据:
struct RtpPacket
{char header[4];struct RtpHeader rtpHeader;uint8_t payload[0];
};
在发送RTP包前,是这样填写这4个字节:
rtpPacket->header[0] = '$';
rtpPacket->header[1] = rtpChannel;
rtpPacket->header[2] = ((dataSize+RTP_HEADER_SIZE) & 0xFF00 ) >> 8;
rtpPacket->header[3] = (dataSize+RTP_HEADER_SIZE) & 0xFF;
🎄三、实现步骤、实现细节
本文与之前的文章【RTSP从零实践】03、实现最简单的传输H264的RTSP服务器 有许多共同点,特点是代码,就是在这个基础去修改的,读者可以比较学习。
使用TCP协议发送RTP包的RTSP服务端实现步骤主要有下面几个:
- 1、创建TCP服务端套接字;
- 2、处理RTSP客户端发过来的命令;
- 3、读取H264文件;
- 4、封装成RTP包并发送
✨3.1 创建TCP服务端套接字
由于RTP包也是使用TCP套接字发送的,所以只需要创建一个TCP套接字即可,不需要再创建UDP套接字了。
int server_fd, client_fd;
struct sockaddr_in address;
int opt = 1;
socklen_t addrlen = sizeof(address);// 创建套接字
if ((server_fd = socket(AF_INET, SOCK_STREAM, 0)) == 0)
{perror("socket failed");return -1;
}// 设置套接字选项
if (setsockopt(server_fd, SOL_SOCKET, SO_REUSEADDR, &opt, sizeof(opt)))
{perror("setsockopt");return -1;
}address.sin_family = AF_INET;
address.sin_addr.s_addr = INADDR_ANY;
address.sin_port = htons(RTSP_PORT);// 绑定端口
if (bind(server_fd, (struct sockaddr *)&address, sizeof(address)) < 0)
{perror("bind failed");return -1;
}// 开始监听
if (listen(server_fd, MAX_CLIENTS) < 0)
{perror("listen");return -1;
}
✨3.2 处理RTSP命令
-
OPTION
-
DESCRIBE
-
SETUP
RTP/AVP/TCP
表示RTP流通过TCP传输,当此值出现时,报文没有client_port字段;
interleaved=0-
1表示streamid,标识RTP的streamid=0;RTCP的streamid=1;
-
PLAY
-
TEARDOWN
✨3.3 读取H264文件
H.264文件保存了h264编码的视频帧,每个视频帧之间以开始码00 00 01
或00 00 00 01
分隔开。我们可以用下面代码判断是否为开始码。
在两个开始码之间的就是视频帧数据。h264视频帧数据的第一个字节是一个NAL头,内容如下图:
可以用下面代码读取NAL头:
以上H264文件结构需要了解的知识,完整代码可以到下个小节查看。
✨3.4 封装成RTP包并发送
本文介绍的RTP数据包是使用TCP协议发送的,主要由3部分组成,即 开始的4个字节、RTP头、RTP负载。
🎈3.4.1、RTP包最开始的4个字节
RTP包最前面的4个字节在前文介绍过了,可以按照下面代码填写:
rtpPacket->header[0] = '$';
rtpPacket->header[1] = rtpChannel;
rtpPacket->header[2] = ((dataSize+RTP_HEADER_SIZE) & 0xFF00 ) >> 8;
rtpPacket->header[3] = (dataSize+RTP_HEADER_SIZE) & 0xFF;
🎈3.4.2、RTP头(RTP Header)
上图是RTP头的结构图,包含了12个字节的内容,可以用代码定义成如下结构体:
struct RtpHeader
{/* byte 0 */uint8_t csrcLen:4;uint8_t extension:1;uint8_t padding:1;uint8_t version:2;/* byte 1 */uint8_t payloadType:7;uint8_t marker:1;/* bytes 2,3 */uint16_t seq;/* bytes 4-7 */uint32_t timestamp;/* bytes 8-11 */uint32_t ssrc;
};
RTP头这里涉及到一个 时间戳怎么计算 的问题,需要注意的是,这个时间戳是一个 时钟频率 为单位的,而不是具体的时间(秒、毫秒等)。
一般情况下,H264的时钟频率为90000Hz,假设帧率为25,那么每一帧的 时间间隔 就是1/25秒,每一帧的 时钟增量 就是(90000/25=3600)。那时间戳怎么算呢?举个例子,如果帧率为25的H264视频,第一帧的RTP时间戳为0的话,那么第二帧的RTP时间戳就是 3600,第三帧的RTP时间戳就是 7200,依次类推,后一帧的RTP时间戳在前一帧的RTP时间戳的值加上一个时钟增量。
🎈3.4.3、H264 的 RTP负载(RTP Payload)
H264 的 RTP负载需要介绍两种方式,第一种是 单个NAL单元封包(Single NAL Unit Packet
);第二种是 分片单元(Fragmentation Unit
) 。如果H264的视频帧NALU(NAL Unit)总字节数小于 MTU(网络最大传输单元1500字节),就可以使用第一种方式,因为有一些TCP/UDP头数据,所以一般判断小于1400字节,就采用 单个NAL单元封包(Single NAL Unit Packet
),否则使用分片单元(Fragmentation Unit
)的方式封装RTP包。
单个NAL单元封包 的RTP负载结构如下图,相当于直接将整个NAL Unit 填入RTP负载即可:
分片单元的RTP负载方式也有两种,本文介绍的是FU-A的方式,RTP负载最开始由三部分组成:第一个字节是FU indicator,第二个字节是FU header,第三个字节开始就是NAL单元去掉NAL头之后的数据:
-
FU indicator
:FU indicator
的大小是一个字节,格式如下,跟NAL头的格式一样,但作为 分片RTP封包 ,并不能直接将H264的NAL头直接填上去。
F
:一般为0。为0表示此NAL单元不应包含bit错误或语法违规;为1表示此NAL单元可能包含bit错误或语法违规;
NRI
:直接将H264NAL头的NRI值填入即可;
Type
:FU-A格式的封包填28
,FU-B格式的封包填29。+---------------+ |0|1|2|3|4|5|6|7| +-+-+-+-+-+-+-+-+ |F|NRI| Type | +---------------+
-
FU header
:FU header
的大小也是一个字节,格式如下:
S
:start,NALU拆分多个分包后,第一个发送的分包,此bit位置1,其他分包为0;
E
:end,NALU拆分多个分包后,最后一个发送的分包,此bit位置1,其他分包为0;
R
:保留位,必须等于0;
Type
:将H264的NAL头的负载类型Type直接填入。+---------------+ |0|1|2|3|4|5|6|7| +-+-+-+-+-+-+-+-+ |S|E|R| Type | +---------------+
这部分可以结合下个小节代码,多看几篇去理解。
🎄四、RTP_over_TCP的RTSP服务端源码
1、H264Reader.h
/*** @file H264Reader.h* @author https://blog.csdn.net/wkd_007* @brief* @version 0.1* @date 2025-06-24** @copyright Copyright (c) 2025**/
#ifndef __H264_READER_H__
#define __H264_READER_H__#include <stdio.h>#define MAX_STARTCODE_LEN (4)typedef enum
{FALSE,TRUE,
} BOOL;typedef enum
{H264_NALU_TYPE_SLICE = 1,H264_NALU_TYPE_DPA = 2,H264_NALU_TYPE_DPB = 3,H264_NALU_TYPE_DPC = 4,H264_NALU_TYPE_IDR = 5,H264_NALU_TYPE_SEI = 6,H264_NALU_TYPE_SPS = 7,H264_NALU_TYPE_PPS = 8,H264_NALU_TYPE_AUD = 9,H264_NALU_TYPE_EOSEQ = 10,H264_NALU_TYPE_EOSTREAM = 11,H264_NALU_TYPE_FILL = 12,
} H264NaluType;typedef enum
{H264_NALU_PRIORITY_DISPOSABLE = 0,H264_NALU_PRIRITY_LOW = 1,H264_NALU_PRIORITY_HIGH = 2,H264_NALU_PRIORITY_HIGHEST = 3
} H264NaluPriority;typedef struct
{int startcode_len; //! 4 for parameter sets and first slice in picture, 3 for everything else (suggested)int forbidden_bit; //! should be always FALSEint nal_reference_idc; //! H264_NALU_PRIORITY_xxxxint nal_unit_type; //! H264_NALU_TYPE_xxxxBOOL isLastFrame; //!int frame_len; //!unsigned char *pFrameBuf; //!
} H264Frame_t;typedef struct H264ReaderInfo_s
{FILE *pFileFd;int frameNum;
} H264ReaderInfo_t;int H264_FileOpen(char *fileName, H264ReaderInfo_t *pH264Info);
int H264_FileClose(H264ReaderInfo_t *pH264Info);
int H264_GetFrame(H264Frame_t *pH264Frame, H264ReaderInfo_t *pH264Info);
BOOL H264_IsEndOfFile(const H264ReaderInfo_t *pH264Info);
void H264_SeekFile(H264ReaderInfo_t *pH264Info);#endif // __H264_READER_H__
2、H264Reader.c
/*** @file H264Reader.c* @author https://blog.csdn.net/wkd_007* @brief* @version 0.1* @date 2025-06-30** @copyright Copyright (c) 2025**/
#include "H264Reader.h"
#include <stdlib.h>#define MAX_FRAME_LEN (1920 * 1080 * 1.5) // 一帧数据最大字节数static BOOL findStartCode_001(unsigned char *Buf)
{// printf("[%d %d %d]\n", Buf[0], Buf[1], Buf[2]);return (Buf[0] == 0 && Buf[1] == 0 && Buf[2] == 1); // 0x000001 ?
}static BOOL findStartCode_0001(unsigned char *Buf)
{// printf("[%d %d %d %d]\n", Buf[0], Buf[1], Buf[2], Buf[3]);return (Buf[0] == 0 && Buf[1] == 0 && Buf[2] == 0 && Buf[3] == 1); // 0x00000001 ?
}int H264_FileOpen(char *fileName, H264ReaderInfo_t *pH264Info)
{pH264Info->pFileFd = fopen(fileName, "rb+");if (pH264Info->pFileFd == NULL){printf("[%s %d]Open file error\n", __FILE__, __LINE__);return -1;}pH264Info->frameNum = 0;return 0;
}int H264_FileClose(H264ReaderInfo_t *pH264Info)
{if (pH264Info->pFileFd != NULL){fclose(pH264Info->pFileFd);pH264Info->pFileFd = NULL;}return 0;
}BOOL H264_IsEndOfFile(const H264ReaderInfo_t *pH264Info)
{return feof(pH264Info->pFileFd);
}void H264_SeekFile(H264ReaderInfo_t *pH264Info)
{fseek(pH264Info->pFileFd, 0, SEEK_SET);pH264Info->frameNum = 0;
}/*** @brief 获取一阵h264视频帧** @param pH264Frame :输出参数,使用后 pH264Frame->pFrameBuf 需要free* @param pH264Info :输入参数* @return int*/
int H264_GetFrame(H264Frame_t *pH264Frame, H264ReaderInfo_t *pH264Info)
{int rewind = 0;if (pH264Info->pFileFd == NULL){printf("[%s %d]pFileFd error\n", __FILE__, __LINE__);return -1;}// 1.读取帧数据// unsigned char *pFrame = (unsigned char *)malloc(MAX_FRAME_LEN);unsigned char *pFrame = pH264Frame->pFrameBuf;int readLen = fread(pFrame, 1, MAX_FRAME_LEN, pH264Info->pFileFd);if (readLen <= 0){printf("[%s %d]fread error\n", __FILE__, __LINE__);// free(pFrame);return -1;}// 2.查找当前帧开始码int i = 0;for (; i < readLen - MAX_STARTCODE_LEN; i++){if (!findStartCode_0001(&pFrame[i])){if (!findStartCode_001(&pFrame[i])){continue;}else{pH264Frame->startcode_len = 3;break;}}else{pH264Frame->startcode_len = 4;break;}}if (i != 0) // 不是帧开头,偏移到帧开头重新读{printf("[%s %d]startcode error, i=%d\n", __FILE__, __LINE__, i);// free(pFrame);rewind = (-(readLen - i));fseek(pH264Info->pFileFd, rewind, SEEK_CUR);return -1;}// 3.查找下一帧开始码i += MAX_STARTCODE_LEN;for (; i < readLen - MAX_STARTCODE_LEN; i++){if (!findStartCode_0001(&pFrame[i])){if (!findStartCode_001(&pFrame[i])){continue;}else{break;}}else{break;}}if (i == (readLen - MAX_STARTCODE_LEN)){if (!feof(pH264Info->pFileFd)){printf("[%s %d]MAX_FRAME_LEN too small\n", __FILE__, __LINE__);// free(pFrame);return -1;}else{pH264Frame->isLastFrame = TRUE;}}// 4.填数据pH264Frame->forbidden_bit = pFrame[pH264Frame->startcode_len] & 0x80; // 1 bitpH264Frame->nal_reference_idc = pFrame[pH264Frame->startcode_len] & 0x60; // 2 bitpH264Frame->nal_unit_type = pFrame[pH264Frame->startcode_len] & 0x1f; // 5 bit, naluType 是开始码后一个字节的最后 5 位// pH264Frame->pFrameBuf = pFrame;pH264Frame->frame_len = i;// 5.文件读取指针偏移到下一帧位置rewind = (-(readLen - i));fseek(pH264Info->pFileFd, rewind, SEEK_CUR);pH264Info->frameNum++;return pH264Frame->frame_len;
}
3、tcp_rtp.h
#ifndef _RTP_H_
#define _RTP_H_
#include <stdint.h>#define RTP_VESION 2#define RTP_PAYLOAD_TYPE_H264 96
#define RTP_PAYLOAD_TYPE_AAC 97#define RTP_HEADER_SIZE 12
#define RTP_MAX_PKT_SIZE 1400/*** 0 1 2 3* 7 6 5 4 3 2 1 0|7 6 5 4 3 2 1 0|7 6 5 4 3 2 1 0|7 6 5 4 3 2 1 0* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+* |V=2|P|X| CC |M| PT | sequence number |* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+* | timestamp |* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+* | synchronization source (SSRC) identifier |* +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+* | contributing source (CSRC) identifiers |* : .... :* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+**/
struct RtpHeader
{/* byte 0 */uint8_t csrcLen:4;uint8_t extension:1;uint8_t padding:1;uint8_t version:2;/* byte 1 */uint8_t payloadType:7;uint8_t marker:1;/* bytes 2,3 */uint16_t seq;/* bytes 4-7 */uint32_t timestamp;/* bytes 8-11 */uint32_t ssrc;
};struct RtpPacket
{char header[4];struct RtpHeader rtpHeader;uint8_t payload[0];
};void rtpHeaderInit(struct RtpPacket* rtpPacket, uint8_t csrcLen, uint8_t extension,uint8_t padding, uint8_t version, uint8_t payloadType, uint8_t marker,uint16_t seq, uint32_t timestamp, uint32_t ssrc);
int rtpSendPacket(int socket, uint8_t rtpChannel, struct RtpPacket* rtpPacket, uint32_t dataSize);#endif //_RTP_H_
4、tcp_rtp.c
#include <sys/types.h>
#include <sys/socket.h>
#include <arpa/inet.h>
#include <netinet/in.h>
#include <arpa/inet.h>#include "tcp_rtp.h"void rtpHeaderInit(struct RtpPacket* rtpPacket, uint8_t csrcLen, uint8_t extension,uint8_t padding, uint8_t version, uint8_t payloadType, uint8_t marker,uint16_t seq, uint32_t timestamp, uint32_t ssrc)
{rtpPacket->rtpHeader.csrcLen = csrcLen;rtpPacket->rtpHeader.extension = extension;rtpPacket->rtpHeader.padding = padding;rtpPacket->rtpHeader.version = version;rtpPacket->rtpHeader.payloadType = payloadType;rtpPacket->rtpHeader.marker = marker;rtpPacket->rtpHeader.seq = seq;rtpPacket->rtpHeader.timestamp = timestamp;rtpPacket->rtpHeader.ssrc = ssrc;
}int rtpSendPacket(int socket, uint8_t rtpChannel, struct RtpPacket* rtpPacket, uint32_t dataSize)
{int ret;rtpPacket->header[0] = '$';rtpPacket->header[1] = rtpChannel;rtpPacket->header[2] = ((dataSize+RTP_HEADER_SIZE) & 0xFF00 ) >> 8;rtpPacket->header[3] = (dataSize+RTP_HEADER_SIZE) & 0xFF;rtpPacket->rtpHeader.seq = htons(rtpPacket->rtpHeader.seq);rtpPacket->rtpHeader.timestamp = htonl(rtpPacket->rtpHeader.timestamp);rtpPacket->rtpHeader.ssrc = htonl(rtpPacket->rtpHeader.ssrc);ret = send(socket, (void*)rtpPacket, dataSize+RTP_HEADER_SIZE+4, 0);rtpPacket->rtpHeader.seq = ntohs(rtpPacket->rtpHeader.seq);rtpPacket->rtpHeader.timestamp = ntohl(rtpPacket->rtpHeader.timestamp);rtpPacket->rtpHeader.ssrc = ntohl(rtpPacket->rtpHeader.ssrc);return ret;
}
5、rtsp_h264_tcp_main.c
/*** @file rtsp_h264_tcp_main.c* @author https://blog.csdn.net/wkd_007* @brief * @version 0.1* @date 2025-07-10* * @copyright Copyright (c) 2025* */
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <fcntl.h>#include "tcp_rtp.h"
#include "H264Reader.h"#define H264_FILE_NAME "test.h264"
#define FPS 25#define RTSP_PORT 8554
#define MAX_CLIENTS 5
#define SESSION_ID 10086001
#define SESSION_TIMEOUT 60typedef struct
{int rtpSendFd;int rtpPort;int rtpChannel;int bPlayFlag; // 播放标志char *cliIp;
} RTP_Send_t;typedef enum
{RTP_NULL,RTP_PLAY,RTP_PLAYING,RTP_STOP,
} RTP_PLAY_STATE;static int rtpSendH264Frame(int socket, int rtpChannel,struct RtpPacket *rtpPacket, uint8_t *frame, uint32_t frameSize)
{uint8_t naluType; // nalu第一个字节int sendBytes = 0;int ret;naluType = frame[0];if (frameSize <= RTP_MAX_PKT_SIZE) // nalu长度小于最大包场:单一NALU单元模式{/** 0 1 2 3 4 5 6 7 8 9* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+* |F|NRI| Type | a single NAL unit ... |* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+*/memcpy(rtpPacket->payload, frame, frameSize);ret = rtpSendPacket(socket, rtpChannel, rtpPacket, frameSize);if (ret < 0)return -1;rtpPacket->rtpHeader.seq++;sendBytes += ret;if ((naluType & 0x1F) == 7 || (naluType & 0x1F) == 8) // 如果是SPS、PPS就不需要加时间戳goto out;}else // nalu长度大于最大包场:分片模式{/** 0 1 2* 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+* | FU indicator | FU header | FU payload ... |* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+*//** FU Indicator* 0 1 2 3 4 5 6 7* +-+-+-+-+-+-+-+-+* |F|NRI| Type |* +---------------+*//** FU Header* 0 1 2 3 4 5 6 7* +-+-+-+-+-+-+-+-+* |S|E|R| Type |* +---------------+*/int pktNum = frameSize / RTP_MAX_PKT_SIZE; // 有几个完整的包int remainPktSize = frameSize % RTP_MAX_PKT_SIZE; // 剩余不完整包的大小int i, pos = 1;/* 发送完整的包 */for (i = 0; i < pktNum; i++){rtpPacket->payload[0] = (naluType & 0x60) | 28;rtpPacket->payload[1] = naluType & 0x1F;if (i == 0) // 第一包数据rtpPacket->payload[1] |= 0x80; // startelse if (remainPktSize == 0 && i == pktNum - 1) // 最后一包数据rtpPacket->payload[1] |= 0x40; // endmemcpy(rtpPacket->payload + 2, frame + pos, RTP_MAX_PKT_SIZE);ret = rtpSendPacket(socket, rtpChannel, rtpPacket, RTP_MAX_PKT_SIZE + 2);if (ret < 0)return -1;rtpPacket->rtpHeader.seq++;sendBytes += ret;pos += RTP_MAX_PKT_SIZE;}/* 发送剩余的数据 */if (remainPktSize > 0){rtpPacket->payload[0] = (naluType & 0x60) | 28;rtpPacket->payload[1] = naluType & 0x1F;rtpPacket->payload[1] |= 0x40; // endmemcpy(rtpPacket->payload + 2, frame + pos, remainPktSize + 2);ret = rtpSendPacket(socket, rtpChannel, rtpPacket, remainPktSize + 2);if (ret < 0)return -1;rtpPacket->rtpHeader.seq++;sendBytes += ret;}}out:return sendBytes;
}void *sendRtp(void *arg)
{RTP_Send_t *pRtpSend = (RTP_Send_t *)arg;int rtp_send_fd = pRtpSend->rtpSendFd;int rtpChannel = pRtpSend->rtpChannel;struct RtpPacket *rtpPacket = (struct RtpPacket *)malloc(sizeof(struct RtpPacket) + (1920 * 1080 * 4));rtpHeaderInit(rtpPacket, 0, 0, 0, RTP_VESION, RTP_PAYLOAD_TYPE_H264, 0,0, 0, 0x88923423);// h264H264ReaderInfo_t h264Info;if (H264_FileOpen(H264_FILE_NAME, &h264Info) < 0){printf("failed to open %s\n", H264_FILE_NAME);return NULL;}H264Frame_t h264Frame;h264Frame.pFrameBuf = (unsigned char *)malloc(1920 * 1080 * 4);while (pRtpSend->bPlayFlag){if (!H264_IsEndOfFile(&h264Info)){h264Frame.isLastFrame = 0;H264_GetFrame(&h264Frame, &h264Info);if (h264Frame.pFrameBuf != NULL){if (h264Frame.isLastFrame) // 最后一帧,移到开头重新读{printf("warning SeekFile 1\n");H264_SeekFile(&h264Info);}// printf("rtpSendH264Frame, frameNum=%d, time=%u\n", h264Info.frameNum, rtpPacket->rtpHeader.timestamp);rtpSendH264Frame(rtp_send_fd, rtpChannel, rtpPacket,h264Frame.pFrameBuf + h264Frame.startcode_len,h264Frame.frame_len - h264Frame.startcode_len);rtpPacket->rtpHeader.timestamp += 90000 / FPS; // RTP 传输视频每秒 90k HZusleep(1000 * 1000 / FPS);}}else{printf("warning need SeekFile 1\n");}}free(h264Frame.pFrameBuf);free(rtpPacket);H264_FileClose(&h264Info);
}// 解析RTSP请求
void rtsp_request_parse(char *buffer, char *method, char *url, int *cseq, int *pRtpChannel)
{char *line = strtok(buffer, "\r\n");sscanf(line, "%s %s RTSP/1.0", method, url);while ((line = strtok(NULL, "\r\n")) != NULL){if (strncmp(line, "CSeq:", 5) == 0){sscanf(line, "CSeq: %d", cseq);}char *pInterleaved = strstr(line, "interleaved=");if (pInterleaved != NULL){int rtcpChn = 0;sscanf(pInterleaved, "interleaved=%d-%d", pRtpChannel, &rtcpChn);// printf("rtpPort: %d-%d\n",*pRtpChannel, rtcpChn);}}
}// 生成SDP描述
const char *generate_sdp()
{return "v=0\r\n""o=- 0 0 IN IP4 0.0.0.0\r\n""s=Example Stream\r\n""t=0 0\r\n""m=video 0 RTP/AVP 96\r\n""a=rtpmap:96 H264/90000\r\n""a=control:streamid=0\r\n";
}void rtsp_handle_OPTION(char *response, int cseq)
{sprintf(response,"RTSP/1.0 200 OK\r\n""CSeq: %d\r\n""Public: OPTIONS, DESCRIBE, SETUP, PLAY, TEARDOWN\r\n\r\n",cseq);
}static void rtsp_handle_DESCRIBE(char *response, int cseq)
{sprintf(response,"RTSP/1.0 200 OK\r\n""CSeq: %d\r\n""Content-Type: application/sdp\r\n""Content-Length: %zu\r\n\r\n%s",cseq, strlen(generate_sdp()), generate_sdp());
}static void rtsp_handle_SETUP(char *response, int cseq, uint8_t rtpChannel)
{sprintf(response,"RTSP/1.0 200 OK\r\n""CSeq: %d\r\n""Session: %u; timeout=%d\r\n""Transport: RTP/AVP/TCP;unicast;interleaved=%hhu-%hhu\r\n\r\n",cseq, SESSION_ID, SESSION_TIMEOUT, rtpChannel, rtpChannel + 1);
}static void rtsp_handle_PLAY(char *response, int cseq)
{sprintf(response,"RTSP/1.0 200 OK\r\n""CSeq: %d\r\n""Session: %u; timeout=%d\r\n""Range: npt=0.000-\r\n\r\n",cseq, SESSION_ID, SESSION_TIMEOUT);
}static void rtsp_handle_TEARDOWN(char *response, int cseq)
{sprintf(response,"RTSP/1.0 200 OK\r\n""CSeq: %d\r\n""Session: %d; timeout=%d\r\n\r\n",cseq, SESSION_ID, SESSION_TIMEOUT);
}// 处理客户端连接
int handle_client(int cli_fd, char *cli_ip)
{int client_sock = cli_fd;char buffer[1024] = {0};int cseq = 0;int rtpChn = 0;unsigned char bSendFlag = RTP_NULL;RTP_Send_t rtpSend;pthread_t thread_id;while (1){memset(buffer, 0, sizeof(buffer));int len = read(client_sock, buffer, sizeof(buffer) - 1);if (len <= 0)break;printf("C->S [\n%s]\n\n", buffer);char method[16] = {0};char url[128] = {0};rtsp_request_parse(buffer, method, url, &cseq, &rtpChn);char response[1024] = {0}; // 构造响应if (strcmp(method, "OPTIONS") == 0){rtsp_handle_OPTION(response, cseq);}else if (strcmp(method, "DESCRIBE") == 0){rtsp_handle_DESCRIBE(response, cseq);}else if (strcmp(method, "SETUP") == 0){rtsp_handle_SETUP(response, cseq, rtpChn);}else if (strcmp(method, "PLAY") == 0){rtsp_handle_PLAY(response, cseq);bSendFlag = RTP_PLAY;}else if (strcmp(method, "TEARDOWN") == 0){rtsp_handle_TEARDOWN(response, cseq);bSendFlag = RTP_STOP;}else{snprintf(response, sizeof(response),"RTSP/1.0 501 Not Implemented\r\nCSeq: %d\r\n\r\n", cseq);}write(client_sock, response, strlen(response));printf("S->C [\n%s]\n\n", response);if (bSendFlag == RTP_PLAY) // PLAY{rtpSend.rtpSendFd = cli_fd;rtpSend.rtpPort = 0;rtpSend.rtpChannel = rtpChn;rtpSend.cliIp = NULL;rtpSend.bPlayFlag = 1;// 这里不使用线程的话,会一直无法处理 client_sock 发过来的 OPTION 消息,导致播放出问题if (pthread_create(&thread_id, NULL, (void *)sendRtp, (void *)&rtpSend) < 0){perror("pthread_create");}bSendFlag = RTP_PLAYING;}if (bSendFlag == RTP_STOP) // TEARDOWN{rtpSend.bPlayFlag = 0;pthread_join(thread_id); // 等待线程结束bSendFlag = RTP_NULL;break;}}printf("close ip=[%s] fd=[%d]\n", cli_ip, client_sock);close(client_sock);return 0;
}int main()
{int server_fd, client_fd;struct sockaddr_in address;int opt = 1;socklen_t addrlen = sizeof(address);// 创建套接字if ((server_fd = socket(AF_INET, SOCK_STREAM, 0)) == 0){perror("socket failed");return -1;}// 设置套接字选项if (setsockopt(server_fd, SOL_SOCKET, SO_REUSEADDR, &opt, sizeof(opt))){perror("setsockopt");return -1;}address.sin_family = AF_INET;address.sin_addr.s_addr = INADDR_ANY;address.sin_port = htons(RTSP_PORT);// 绑定端口if (bind(server_fd, (struct sockaddr *)&address, sizeof(address)) < 0){perror("bind failed");return -1;}// 开始监听if (listen(server_fd, MAX_CLIENTS) < 0){perror("listen");return -1;}printf("RTSP Server listening on port %d\n", RTSP_PORT);// 主循环接受连接,目前处理一个客户端while (1){char cli_ip[40] = {0};if ((client_fd = accept(server_fd, (struct sockaddr *)&address, &addrlen)) < 0){perror("accept");return -1;}strncpy(cli_ip, inet_ntoa(address.sin_addr), sizeof(cli_ip));printf("handle cliend [%s]\n", cli_ip);handle_client(client_fd, cli_ip);}return 0;
}
首先设置一下VLC,工具->偏好设置->输入/编解码器,勾选如下图的RTP over RTSP(TCP)
将上面代码保存在同一个目录后,并且在同目录里放一个.h264文件,然后运行 gcc *.c -lpthread
编译,再执行./a.out运行程序,下面是我运行的过程:
🎄五、、总结
本文介绍了RTP_over_TCP的一些概念,以及TCP传输H264RTP包的RTSP服务器实现的步骤和细节,最后提供了实现的源代码,帮助读者学习理解。
如果文章有帮助的话,点赞👍、收藏⭐,支持一波,谢谢 😁😁😁
参考:
https://blog.csdn.net/huabiaochen/article/details/104582872