liveMedia/VorbisAudioRTPSink.cpp

Go to the documentation of this file.

/**********
This library is free software; you can redistribute it and/or modify it under
the terms of the GNU Lesser General Public License as published by the
Free Software Foundation; either version 2.1 of the License, or (at your
option) any later version. (See <http://www.gnu.org/copyleft/lesser.html>.)

This library is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License for
more details.

You should have received a copy of the GNU Lesser General Public License
along with this library; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
**********/
// "liveMedia"
// Copyright (c) 1996-2012 Live Networks, Inc.  All rights reserved.
// RTP sink for Vorbis audio
// Implementation

#include "VorbisAudioRTPSink.hh"
#include "Base64.hh"

VorbisAudioRTPSink::VorbisAudioRTPSink(UsageEnvironment& env, Groupsock* RTPgs,
                                       u_int8_t rtpPayloadFormat,
                                       u_int32_t rtpTimestampFrequency,
                                       unsigned numChannels,
                                       u_int8_t* identificationHeader, unsigned identificationHeaderSize,
                                       u_int8_t* commentHeader, unsigned commentHeaderSize,
                                       u_int8_t* setupHeader, unsigned setupHeaderSize,
                                       u_int32_t identField)
  : AudioRTPSink(env, RTPgs, rtpPayloadFormat, rtpTimestampFrequency, "VORBIS", numChannels),
    fIdent(identField), fFmtpSDPLine(NULL) {
  // Create packed configuration headers, and encode this data into a "a=fmtp:" SDP line that we'll use to describe it:
  
  // First, count how many headers (<=3) are included, and how many bytes will be used to encode these headers' sizes:
  unsigned numHeaders = 0;
  unsigned sizeSize[2]; // The number of bytes used to encode the lengths of the first two headers (but not the length of the 3rd)
  sizeSize[0] = sizeSize[1] = 0;
  if (identificationHeaderSize > 0) {
    sizeSize[numHeaders++] = identificationHeaderSize < 128 ? 1 : identificationHeaderSize < 16384 ? 2 : 3;
  }
  if (commentHeaderSize > 0) {
    sizeSize[numHeaders++] = commentHeaderSize < 128 ? 1 : commentHeaderSize < 16384 ? 2 : 3;
  }
  if (setupHeaderSize > 0) {
    ++numHeaders;
  } else {
    sizeSize[1] = 0; // We have at most two headers, so the second one's length isn't encoded
  }
  if (numHeaders == 0) return; // With no headers, we can't set up a configuration
  if (numHeaders == 1) sizeSize[0] = 0; // With only one header, its length isn't encoded

  // Then figure out the size of the packed configuration headers, and allocate space for this:
  unsigned length = identificationHeaderSize + commentHeaderSize + setupHeaderSize; // The "length" field in the packed headers
  if (length > (unsigned)0xFFFF) return; // too big for a 16-bit field; we can't handle this
  unsigned packedHeadersSize
    = 4 // "Number of packed headers" field
    + 3 // "ident" field
    + 2 // "length" field
    + 1 // "n. of headers" field
    + sizeSize[0] + sizeSize[1] // "length1" and "length2" (if present) fields
    + length;
  u_int8_t* packedHeaders = new u_int8_t[packedHeadersSize];
  if (packedHeaders == NULL) return;

  // Fill in the 'packed headers':
  u_int8_t* p = packedHeaders;
  *p++ = 0; *p++ = 0; *p++ = 0; *p++ = 1; // "Number of packed headers": 1
  *p++ = fIdent>>16; *p++ = fIdent>>8; *p++ = fIdent; // "Ident" (24 bits)
  *p++ = length>>8; *p++ = length; // "length" (16 bits)
  *p++ = numHeaders-1; // "n. of headers"
  if (numHeaders > 1) {
    // Fill in the "length1" header:
    unsigned length1 = identificationHeaderSize > 0 ? identificationHeaderSize : commentHeaderSize;
    if (length1 >= 16384) {
      *p++ = 0x80; // flag, but no more, because we know length1 <= 32767
    }
    if (length1 >= 128) {
      *p++ = 0x80|((length1&0x3F80)>>7); // flag + the second 7 bits
    }
    *p++ = length1&0x7F; // the low 7 bits

    if (numHeaders > 2) { // numHeaders == 3
      // Fill in the "length2" header (for the 'Comment' header):
      unsigned length2 = commentHeaderSize;
      if (length2 >= 16384) {
        *p++ = 0x80; // flag, but no more, because we know length2 <= 32767
      }
      if (length2 >= 128) {
        *p++ = 0x80|((length2&0x3F80)>>7); // flag + the second 7 bits
      }
      *p++ = length2&0x7F; // the low 7 bits
    }
  }
  // Copy each header:
  if (identificationHeader != NULL) memmove(p, identificationHeader, identificationHeaderSize); p += identificationHeaderSize;
  if (commentHeader != NULL) memmove(p, commentHeader, commentHeaderSize); p += commentHeaderSize;
  if (setupHeader != NULL) memmove(p, setupHeader, setupHeaderSize);
  
  // Having set up the 'packed configuration headers', Base-64-encode this, and put it in our "a=fmtp:" SDP line:
  char* base64PackedHeaders = base64Encode((char const*)packedHeaders, packedHeadersSize);
  delete[] packedHeaders;
  
  unsigned fmtpSDPLineMaxSize = 50 + strlen(base64PackedHeaders); // 50 => more than enough space
  fFmtpSDPLine = new char[fmtpSDPLineMaxSize];
  sprintf(fFmtpSDPLine, "a=fmtp:%d configuration=%s\r\n", rtpPayloadType(), base64PackedHeaders);
  delete[] base64PackedHeaders;
}

VorbisAudioRTPSink::~VorbisAudioRTPSink() {
  delete[] fFmtpSDPLine;
}

VorbisAudioRTPSink*
VorbisAudioRTPSink::createNew(UsageEnvironment& env, Groupsock* RTPgs,
                              u_int8_t rtpPayloadFormat, u_int32_t rtpTimestampFrequency, unsigned numChannels,
                              u_int8_t* identificationHeader, unsigned identificationHeaderSize,
                              u_int8_t* commentHeader, unsigned commentHeaderSize,
                              u_int8_t* setupHeader, unsigned setupHeaderSize) {
  return new VorbisAudioRTPSink(env, RTPgs,
                                rtpPayloadFormat, rtpTimestampFrequency, numChannels,
                                identificationHeader, identificationHeaderSize,
                                commentHeader, commentHeaderSize,
                                setupHeader, setupHeaderSize);
}

#define ADVANCE(n) do { p += (n); rem -= (n); } while (0)
#define GET_ENCODED_VAL(n) do { u_int8_t byte; n = 0; do { if (rem == 0) break; byte = *p; n = n*128 + byte&0x7F; ADVANCE(1); } while (byte&0x80); } while (0); if (rem == 0) break

VorbisAudioRTPSink*
VorbisAudioRTPSink::createNew(UsageEnvironment& env, Groupsock* RTPgs,
                              u_int8_t rtpPayloadFormat, u_int32_t rtpTimestampFrequency, unsigned numChannels,
                              char const* configStr) {
  u_int8_t* identificationHeader = NULL; unsigned identificationHeaderSize = 0;
  u_int8_t* commentHeader = NULL; unsigned commentHeaderSize = 0;
  u_int8_t* setupHeader = NULL; unsigned setupHeaderSize = 0;
  VorbisAudioRTPSink* resultSink = NULL;

  // Begin by Base64-decoding the configuration string:
  unsigned configDataSize;
  u_int8_t* configData = base64Decode(configStr, configDataSize);
  u_int8_t* p = configData;
  unsigned rem = configDataSize;

  do {
    if (rem < 4) break;
    u_int32_t numPackedHeaders = (p[0]<<24)|(p[1]<<16)|(p[2]<<8)|p[3]; ADVANCE(4);

    if (numPackedHeaders == 0) break;
    // Use the first 'packed header' only.

    if (rem < 3) break;
    u_int32_t ident = (p[0]<<16)|(p[1]<<8)|p[2]; ADVANCE(3);

    if (rem < 2) break;
    u_int16_t length = (p[0]<<8)|p[1]; ADVANCE(2);

    unsigned numHeaders;
    GET_ENCODED_VAL(numHeaders);

    Boolean success = False;
    for (unsigned i = 0; i < numHeaders+1 && i < 3; ++i) {
      success = False;
      unsigned headerSize;
      if (i < numHeaders) {
        // The header size is encoded:
        GET_ENCODED_VAL(headerSize);
        if (headerSize > length) break;
        length -= headerSize;
      } else {
        // The last header is implicit:
        headerSize = length;
      }

      // Allocate space for the header bytes; we'll fill it in later
      if (i == 0) {
        identificationHeaderSize = headerSize;
        identificationHeader = new u_int8_t[identificationHeaderSize];
      } else if (i == 1) {
        commentHeaderSize = headerSize;
        commentHeader = new u_int8_t[commentHeaderSize];
      } else { // i == 2
        setupHeaderSize = headerSize;
        setupHeader = new u_int8_t[setupHeaderSize];
      }

      success = True;
    }
    if (!success) break;

    // Copy the remaining config bytes into the appropriate 'header' buffers:
    if (identificationHeader != NULL) {
      memmove(identificationHeader, p, identificationHeaderSize); ADVANCE(identificationHeaderSize);
      if (commentHeader != NULL) {
        memmove(commentHeader, p, commentHeaderSize); ADVANCE(commentHeaderSize);
        if (setupHeader != NULL) {
          memmove(setupHeader, p, setupHeaderSize); ADVANCE(setupHeaderSize);
        }
      }
    }

    resultSink = new VorbisAudioRTPSink(env, RTPgs, rtpPayloadFormat, rtpTimestampFrequency, numChannels,
                                        identificationHeader, identificationHeaderSize,
                                        commentHeader, commentHeaderSize,
                                        setupHeader, setupHeaderSize,
                                        ident);
  } while (0);

  delete[] configData;
  delete[] identificationHeader; delete[] commentHeader; delete[] setupHeader;
  return resultSink;
}

char const* VorbisAudioRTPSink::auxSDPLine() {
  return fFmtpSDPLine;
}

void VorbisAudioRTPSink
::doSpecialFrameHandling(unsigned fragmentationOffset,
                         unsigned char* frameStart,
                         unsigned numBytesInFrame,
                         struct timeval framePresentationTime,
                         unsigned numRemainingBytes) {
  // Set the 4-byte "payload header", as defined in RFC 5215, section 2.2:
  u_int8_t header[4];

  // The three bytes of the header are our "Ident":
  header[0] = fIdent>>16; header[1] = fIdent>>8; header[2] = fIdent;

  // The final byte contains the "F", "VDT", and "numPkts" fields:
  u_int8_t F; // Fragment type
  if (numRemainingBytes > 0) {
    if (fragmentationOffset > 0) {
      F = 2<<6; // continuation fragment
    } else {
      F = 1<<6; // start fragment
    }
  } else {
    if (fragmentationOffset > 0) {
      F = 3<<6; // end fragment
    } else {
      F = 0<<6; // not fragmented
    }
  }
  u_int8_t const VDT = 0<<4; // Vorbis Data Type (always a "Raw Vorbis payload")
  u_int8_t numPkts = F == 0 ? (numFramesUsedSoFar() + 1): 0; // set to 0 when we're a fragment
  header[3] = F|VDT|numPkts;

  setSpecialHeaderBytes(header, sizeof header);

  // There's also a 2-byte 'frame-specific' header: The length of the Vorbis data:
  u_int8_t frameSpecificHeader[2];
  frameSpecificHeader[0] = numBytesInFrame>>8;
  frameSpecificHeader[1] = numBytesInFrame;
  setFrameSpecificHeaderBytes(frameSpecificHeader, 2);

  // Important: Also call our base class's doSpecialFrameHandling(),
  // to set the packet's timestamp:
  MultiFramedRTPSink::doSpecialFrameHandling(fragmentationOffset,
                                             frameStart, numBytesInFrame,
                                             framePresentationTime,
                                             numRemainingBytes);
}

Boolean VorbisAudioRTPSink::frameCanAppearAfterPacketStart(unsigned char const* /*frameStart*/,
                                                           unsigned /*numBytesInFrame*/) const {
  // We allow more than one frame to be packed into an outgoing RTP packet, but no more than 15:
  return numFramesUsedSoFar() <= 15;
}

unsigned VorbisAudioRTPSink::specialHeaderSize() const {
  return 4;
}

unsigned VorbisAudioRTPSink::frameSpecificHeaderSize() const {
  return 2;
}

---