从RTSP协议SDP数据中获得二进制的SPS、PPS

来源：http://www.w2bc.com/Article/43572

在RTSP协议的交互过程中，第二步客户端发送DESCRIBE请求之后，服务端会返回SDP内容，该SDP内容中有关于媒体和会话的描述，本篇文章主要给出如何从SDP字符串中得到H264视频信息中的sps、pps的二进制数据。

我们知道，在RTSP协议中DESCRIBE请求回复内容的SDP部分中，如果服务端的直播流的视频是H264的编码格式的话，那么在SDP中会将H264的sps、pps信息通过Base64编码成字符串发送给客户端(也就是解码器端)，sps称为序列参数集，pps称为图形参数集。这两个参数中包含了初始化H.264解码器所需要的信息参数，包括编码所用的profile，level，图像的宽和高，deblock滤波器等。这样解码器就可以在DESCRIBE阶段，利用这些参数初始化解码器的设置了。那么如何将SDP中的字符串还原成sps、pps的二进制数据呢。下面的部分代码是从live555项目中取出来的，可以作为小功能独立使用，如果大家有用的着，可以直接拿去使用在项目中：

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

//main.cpp的内容   #include <stdio.h> #include "Base64.h"   int main() {      /*          RTSP 响应的SDP的内容中sprop-parameter-sets键值：          sprop-parameter-sets=Z2QAKq2wpDBSAgFxQWKQPQRWFIYKQEAuKCxSB6CKwpDBSAgFxQWKQPQRTDoUKQNC4oJHMGIemHQoUgaFxQSOYMQ9MOhQpA0LigkcwYh6xEQmIVilsQRWUURJsogxOU4QITKUIEVlCCTYQVhBMJQhMIjGggWQJFaIGBJZBAaEnaMIDwsSWQQKCwsrRBQYOWQweO0YEBZASNAogszlAUAW7/wcFBwMQAABdwAAr8g4AAADAL68IAAAdzWU//+MAAADAF9eEAAAO5rKf//CgA==,aP48sA==;          其中逗号前面的内容是sps的二进制数据被base64之后的结果          而逗号后面的内容(不要分号,分号是sdp中键值对的分隔符),是pps的内容          使用live555中的base64Decode函数分别对这两部分进行反base64解码得到的二进制数据就是h264中的sps pps 的二进制内容          分别是以67 和 68 开头      */      char * sps_sdp = "Z2QAKq2wpDBSAgFxQWKQPQRWFIYKQEAuKCxSB6CKwpDBSAgFxQWKQPQRTDoUKQNC4oJHMGIemHQoUgaFxQSOYMQ9MOhQpA0LigkcwYh6xEQmIVilsQRWUURJsogxOU4QITKUIEVlCCTYQVhBMJQhMIjGggWQJFaIGBJZBAaEnaMIDwsSWQQKCwsrRBQYOWQweO0YEBZASNAogszlAUAW7/wcFBwMQAABdwAAr8g4AAADAL68IAAAdzWU//+MAAADAF9eEAAAO5rKf//CgA==" ;      char * pps_sdp = "aP48sA==" ;      unsigned int result_size=0;      unsigned char * p = base64Decode(sps_sdp,result_size);      for ( int i =0;i<result_size;i++)      {          printf ( "%02X " ,p[i]);          if ((i+1)%16==0)          {              printf ( "\n" );          }            }      printf ( "\n\n\n" );        p = base64Decode(pps_sdp,result_size);      for ( int i =0;i<result_size;i++)      {          printf ( "%02X " ,p[i]);          if ((i+1)%16==0)          {              printf ( "\n" );          }            }      printf ( "\n" );        return 0 ; }<br> /* 程序的解码输出如下,得到的分别是3500的sps和pps内容：   67 64 00 2A AD B0 A4 30 52 02 01 71 41 62 90 3D 04 56 14 86 0A 40 40 2E 28 2C 52 07 A0 8A C2 90 C1 48 08 05 C5 05 8A 40 F4 11 4C 3A 14 29 03 42 E2 82 47 30 62 1E 98 74 28 52 06 85 C5 04 8E 60 C4 3D 30 E8 50 A4 0D 0B 8A 09 1C C1 88 7A C4 44 26 21 58 A5 B1 04 56 51 44 49 B2 88 31 39 4E 10 21 32 94 20 45 65 08 24 D8 41 58 41 30 94 21 30 88 C6 82 05 90 24 56 88 18 12 59 04 06 84 9D A3 08 0F 0B 12 59 04 0A 0B 0B 2B 44 14 18 39 64 30 78 ED 18 10 16 40 48 D0 28 82 CC E5 01 40 16 EF FC 1C 14 1C 0C 40 00 01 77 00 00 AF C8 38 00 00 03 00 BE BC 20 00 00 77 35 94 FF FF 8C 00 00 03 00 5F 5E 10 00 00 3B 9A CA 7F FF C2 80   68 FE 3C B0 */

 

 其中用到的一个主要函数 base64Decode 的实现如下：

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93

#include "Base64.h" #include "strDup.h" #include <string.h>   static char base64DecodeTable[256];   static void initBase64DecodeTable() {    int i;    for (i = 0; i < 256; ++i) base64DecodeTable[i] = ( char )0x80;        // default value: invalid      for (i = 'A' ; i <= 'Z' ; ++i) base64DecodeTable[i] = 0 + (i - 'A' );    for (i = 'a' ; i <= 'z' ; ++i) base64DecodeTable[i] = 26 + (i - 'a' );    for (i = '0' ; i <= '9' ; ++i) base64DecodeTable[i] = 52 + (i - '0' );    base64DecodeTable[(unsigned char ) '+' ] = 62;    base64DecodeTable[(unsigned char ) '/' ] = 63;    base64DecodeTable[(unsigned char ) '=' ] = 0; }   unsigned char * base64Decode( char const * in, unsigned& resultSize,                  Boolean trimTrailingZeros) {    static Boolean haveInitedBase64DecodeTable = False;    if (!haveInitedBase64DecodeTable) {      initBase64DecodeTable();      haveInitedBase64DecodeTable = True;    }      unsigned char * out = (unsigned char *)strDupSize(in); // ensures we have enough space    int k = 0;    int const jMax = strlen (in) - 3;       // in case "in" is not a multiple of 4 bytes (although it should be)    for ( int j = 0; j < jMax; j += 4) {      char inTmp[4], outTmp[4];      for ( int i = 0; i < 4; ++i) {        inTmp[i] = in[i+j];        outTmp[i] = base64DecodeTable[(unsigned char )inTmp[i]];        if ((outTmp[i]&0x80) != 0) outTmp[i] = 0; // pretend the input was 'A'      }        out[k++] = (outTmp[0]<<2) | (outTmp[1]>>4);      out[k++] = (outTmp[1]<<4) | (outTmp[2]>>2);      out[k++] = (outTmp[2]<<6) | outTmp[3];    }      if (trimTrailingZeros) {      while (k > 0 && out[k-1] == '\0' ) --k;    }    resultSize = k;    unsigned char * result = new unsigned char [resultSize];    memmove (result, out, resultSize);    delete [] out;      return result; }   static const char base64Char[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/" ;   char * base64Encode( char const * origSigned, unsigned origLength) {    unsigned char const * orig = (unsigned char const *)origSigned; // in case any input bytes have the MSB set    if (orig == NULL) return NULL;      unsigned const numOrig24BitValues = origLength/3;    Boolean havePadding = origLength > numOrig24BitValues*3;    Boolean havePadding2 = origLength == numOrig24BitValues*3 + 2;    unsigned const numResultBytes = 4*(numOrig24BitValues + havePadding);    char * result = new char [numResultBytes+1]; // allow for trailing '\0'      // Map each full group of 3 input bytes into 4 output base-64 characters:    unsigned i;    for (i = 0; i < numOrig24BitValues; ++i) {      result[4*i+0] = base64Char[(orig[3*i]>>2)&0x3F];      result[4*i+1] = base64Char[(((orig[3*i]&0x3)<<4) | (orig[3*i+1]>>4))&0x3F];      result[4*i+2] = base64Char[((orig[3*i+1]<<2) | (orig[3*i+2]>>6))&0x3F];      result[4*i+3] = base64Char[orig[3*i+2]&0x3F];    }      // Now, take padding into account.  (Note: i == numOrig24BitValues)    if (havePadding) {      result[4*i+0] = base64Char[(orig[3*i]>>2)&0x3F];      if (havePadding2) {        result[4*i+1] = base64Char[(((orig[3*i]&0x3)<<4) | (orig[3*i+1]>>4))&0x3F];        result[4*i+2] = base64Char[(orig[3*i+1]<<2)&0x3F];      } else {        result[4*i+1] = base64Char[((orig[3*i]&0x3)<<4)&0x3F];        result[4*i+2] = '=' ;      }      result[4*i+3] = '=' ;    }      result[numResultBytes] = '\0' ;    return result; }

 完整demo下载(Ubuntu Linux下运行没问题)