audio.c

这个库实现了录象功能

/*******************************************************************************
 audio.c

 libquicktime - A library for reading and writing quicktime/avi/mp4 files.
 http://libquicktime.sourceforge.net

 Copyright (C) 2002 Heroine Virtual Ltd.
 Copyright (C) 2002-2007 Members of the libquicktime project.

 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.

 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
*******************************************************************************/ 

#include "lqt_private.h"
#include "ffmpeg.h"
#define LQT_LIBQUICKTIME
#include <quicktime/lqt_codecapi.h>
#include <stdlib.h>
#include <string.h>

#define LOG_DOMAIN "ffmpeg_audio"

/* Different decoding functions */

#if LIBAVCODEC_BUILD >= 3349760
#define DECODE_FUNC avcodec_decode_audio2
#else
#define DECODE_FUNC avcodec_decode_audio
#endif

/* The following code was ported from gmerlin_avdecoder (http://gmerlin.sourceforge.net) */

/* MPEG Audio header parsing code */

static int mpeg_bitrates[5][16] = {
  /* MPEG-1 */
  { 0,  32000,  64000,  96000, 128000, 160000, 192000, 224000,    // I
       256000, 288000, 320000, 352000, 384000, 416000, 448000, 0},
  { 0,  32000,  48000,  56000,  64000,  80000,  96000, 112000,    // II
       128000, 160000, 192000, 224000, 256000, 320000, 384000, 0 },
  { 0,  32000,  40000,  48000,  56000,  64000,  80000,  96000,    // III
       112000, 128000, 160000, 192000, 224000, 256000, 320000, 0 },

  /* MPEG-2 LSF */
  { 0,  32000,  48000,  56000,  64000,  80000,  96000, 112000,    // I
       128000, 144000, 160000, 176000, 192000, 224000, 256000, 0 },
  { 0,   8000,  16000,  24000,  32000,  40000,  48000,  56000,
        64000,  80000,  96000, 112000, 128000, 144000, 160000, 0 } // II & III
};

static int mpeg_samplerates[3][3] = {
  { 44100, 48000, 32000 }, // MPEG1
  { 22050, 24000, 16000 }, // MPEG2
  { 11025, 12000, 8000 }   // MPEG2.5
  };

#define MPEG_ID_MASK        0x00180000
#define MPEG_MPEG1          0x00180000
#define MPEG_MPEG2          0x00100000
#define MPEG_MPEG2_5        0x00000000

#define MPEG_LAYER_MASK     0x00060000
#define MPEG_LAYER_III      0x00020000
#define MPEG_LAYER_II       0x00040000
#define MPEG_LAYER_I        0x00060000
#define MPEG_PROTECTION     0x00010000
#define MPEG_BITRATE_MASK   0x0000F000
#define MPEG_FREQUENCY_MASK 0x00000C00
#define MPEG_PAD_MASK       0x00000200
#define MPEG_PRIVATE_MASK   0x00000100
#define MPEG_MODE_MASK      0x000000C0
#define MPEG_MODE_EXT_MASK  0x00000030
#define MPEG_COPYRIGHT_MASK 0x00000008
#define MPEG_HOME_MASK      0x00000004
#define MPEG_EMPHASIS_MASK  0x00000003
#define LAYER_I_SAMPLES       384
#define LAYER_II_III_SAMPLES 1152

/* Header detection stolen from the mpg123 plugin of xmms */

static int header_check(uint32_t head)
{
        if ((head & 0xffe00000) != 0xffe00000)
                return 0;
        if (!((head >> 17) & 3))
                return 0;
        if (((head >> 12) & 0xf) == 0xf)
                return 0;
        if (!((head >> 12) & 0xf))
                return 0;
        if (((head >> 10) & 0x3) == 0x3)
                return 0;
        if (((head >> 19) & 1) == 1 &&
            ((head >> 17) & 3) == 3 &&
            ((head >> 16) & 1) == 1)
                return 0;
        if ((head & 0xffff0000) == 0xfffe0000)
                return 0;
        return 1;
}

typedef enum
  {
    MPEG_VERSION_NONE = 0,
    MPEG_VERSION_1 = 1,
    MPEG_VERSION_2 = 2,
    MPEG_VERSION_2_5
  } mpeg_version_t;

#define CHANNEL_STEREO   0
#define CHANNEL_JSTEREO  1
#define CHANNEL_DUAL     2
#define CHANNEL_MONO     3

typedef struct
  {
  mpeg_version_t version;
  int layer;
  int bitrate;    /* -1: VBR */
  int samplerate;
  int frame_bytes;
  int channel_mode;
  int mode;
  int samples_per_frame;
  } mpeg_header;

static int header_equal(const mpeg_header * h1, const mpeg_header * h2)
  {
  return ((h1->layer == h2->layer) && (h1->version == h2->version) &&
          (h1->samplerate == h2->samplerate));
  }

static int decode_header(mpeg_header * h, uint8_t * ptr, const mpeg_header * ref)
  {
  uint32_t header;
  int index;
  /* For calculation of the byte length of a frame */
  int pad;
  int slots_per_frame;
  h->frame_bytes = 0;
  header =
    ptr[3] | (ptr[2] << 8) | (ptr[1] << 16) | (ptr[0] << 24);
  if(!header_check(header))
    return 0;
  index = (header & MPEG_MODE_MASK) >> 6;
  switch(index)
    {
    case 0:
      h->channel_mode = CHANNEL_STEREO;
      break;
    case 1:
      h->channel_mode = CHANNEL_JSTEREO;
      break;
    case 2:
      h->channel_mode = CHANNEL_DUAL;
      break;
    case 3:
      h->channel_mode = CHANNEL_MONO;
      break;
    }
  /* Get Version */
  switch(header & MPEG_ID_MASK)
    {
    case MPEG_MPEG1:
      h->version = MPEG_VERSION_1;
        break;
    case MPEG_MPEG2:
      h->version = MPEG_VERSION_2;
      break;
    case MPEG_MPEG2_5:
      h->version = MPEG_VERSION_2_5;
      break;
    default:
      return 0;
    }
  /* Get Layer */
  switch(header & MPEG_LAYER_MASK)
    {
    case MPEG_LAYER_I:
      h->layer = 1;
      break;
    case MPEG_LAYER_II:
      h->layer = 2;
      break;
    case MPEG_LAYER_III:
      h->layer = 3;
      break;
    }
  index = (header & MPEG_BITRATE_MASK) >> 12;
  switch(h->version)
    {
    case MPEG_VERSION_1:
      switch(h->layer)
        {
        case 1:
          h->bitrate = mpeg_bitrates[0][index];
          break;
        case 2:
          h->bitrate = mpeg_bitrates[1][index];
          break;
        case 3:
          h->bitrate = mpeg_bitrates[2][index];
          break;
        }
      break;
    case MPEG_VERSION_2:
    case MPEG_VERSION_2_5:
      switch(h->layer)
        {
        case 1:
          h->bitrate = mpeg_bitrates[3][index];
          break;
        case 2:
        case 3:
          h->bitrate = mpeg_bitrates[4][index];
          break;
        }
      break;
    default: // This won't happen, but keeps gcc quiet
      return 0;
    }
  index = (header & MPEG_FREQUENCY_MASK) >> 10;
  switch(h->version)
    {
    case MPEG_VERSION_1:
      h->samplerate = mpeg_samplerates[0][index];
      break;
    case MPEG_VERSION_2:
      h->samplerate = mpeg_samplerates[1][index];
      break;
    case MPEG_VERSION_2_5:
      h->samplerate = mpeg_samplerates[2][index];
      break;
    default: // This won't happen, but keeps gcc quiet
      return 0;
    }
  pad = (header & MPEG_PAD_MASK) ? 1 : 0;
  if(h->layer == 1)
    {
    h->frame_bytes = ((12 * h->bitrate / h->samplerate) + pad) * 4;
    }
  else
    {
    slots_per_frame = ((h->layer == 3) &&
      ((h->version == MPEG_VERSION_2) ||
       (h->version == MPEG_VERSION_2_5))) ? 72 : 144;
    h->frame_bytes = (slots_per_frame * h->bitrate) / h->samplerate + pad;
    }
  // h->mode = (ptr[3] >> 6) & 3;
  h->samples_per_frame =
    (h->layer == 1) ? LAYER_I_SAMPLES : LAYER_II_III_SAMPLES;
  if(h->version != MPEG_VERSION_1)
    h->samples_per_frame /= 2;
  //  dump_header(h);

  /* Check against reference header */

  if(ref && !header_equal(ref, h))
    return 0;
  return 1;
  }

typedef struct
  {
  AVCodecContext * avctx;
  AVCodec * encoder;
  AVCodec * decoder;

  int initialized;
  
  /* Interleaved samples as avcodec needs them */
    
  int16_t * sample_buffer;
  int sample_buffer_alloc; 
  int samples_in_buffer;
  
  /* Buffer for the entire chunk */

  uint8_t * chunk_buffer;
  int chunk_buffer_alloc;
  int bytes_in_chunk_buffer;

  /* Start and end positions of the sample buffer */
    
  int64_t sample_buffer_start;
  int64_t sample_buffer_end;

  mpeg_header mph;
  int have_mpeg_header;

  uint8_t * extradata;
    
  } quicktime_ffmpeg_audio_codec_t;


static int lqt_ffmpeg_delete_audio(quicktime_audio_map_t *vtrack)
  {
  quicktime_ffmpeg_audio_codec_t *codec = ((quicktime_codec_t*)vtrack->codec)->priv;
  
  if(codec->avctx)
    {
    if(codec->initialized)
      avcodec_close(codec->avctx);
    av_free(codec->avctx);
    }
  if(codec->sample_buffer) free(codec->sample_buffer);
  if(codec->chunk_buffer)  free(codec->chunk_buffer);
  if(codec->extradata) free(codec->extradata); 
  free(codec);
  return 0;
  }

static int set_parameter(quicktime_t *file, 
                  int track, 
                  const char *key, 
                  const void *value)
  {
  quicktime_ffmpeg_audio_codec_t *codec = ((quicktime_codec_t*)file->atracks[track].codec)->priv;
  lqt_ffmpeg_set_parameter(codec->avctx, key, value);
  return 0;
  }


/* Decode VBR chunk into the sample buffer */

static int decode_chunk_vbr(quicktime_t * file, int track)
  {
  int chunk_packets, i, num_samples, bytes_decoded;
  int packet_size, packet_samples;
  int frame_bytes;
  int new_samples;
  quicktime_audio_map_t *track_map = &(file->atracks[track]);
  quicktime_ffmpeg_audio_codec_t *codec = ((quicktime_codec_t*)track_map->codec)->priv;

  chunk_packets = lqt_audio_num_vbr_packets(file, track, track_map->current_chunk, &num_samples);

  if(!chunk_packets)
    return 0;

  new_samples = num_samples + AVCODEC_MAX_AUDIO_FRAME_SIZE / (2 * track_map->channels);
  
  if(codec->sample_buffer_alloc <
     codec->sample_buffer_end - codec->sample_buffer_start + new_samples)
    {
    
    codec->sample_buffer_alloc = codec->sample_buffer_end - codec->sample_buffer_start + new_samples;
    codec->sample_buffer = realloc(codec->sample_buffer, 2 * codec->sample_buffer_alloc *
                                   track_map->channels);
    }

  for(i = 0; i < chunk_packets; i++)
    {
    packet_size = lqt_audio_read_vbr_packet(file, track, track_map->current_chunk, i,
                                            &(codec->chunk_buffer), &(codec->chunk_buffer_alloc),
                                            &packet_samples);
    if(!packet_size)
      return 0;

#if LIBAVCODEC_BUILD >= 3349760
    bytes_decoded = codec->sample_buffer_alloc -
      (codec->sample_buffer_end - codec->sample_buffer_start);
    bytes_decoded *= 2 * track_map->channels;
#else
    bytes_decoded = 0;
#endif

    
    frame_bytes =
      DECODE_FUNC(codec->avctx,
                  &(codec->sample_buffer[track_map->channels *
                                         (codec->sample_buffer_end - codec->sample_buffer_start)]),
                  &bytes_decoded,
                  codec->chunk_buffer,
                  packet_size + FF_INPUT_BUFFER_PADDING_SIZE);
    if(frame_bytes < 0)
      {
      lqt_log(file, LQT_LOG_ERROR, LOG_DOMAIN, "avcodec_decode_audio error");
      break;
      }
    codec->sample_buffer_end += (bytes_decoded / (track_map->channels * 2));
    }
  track_map->current_chunk++;
  return num_samples;
  }


/* Decode the current chunk into the sample buffer */

static int decode_chunk(quicktime_t * file, int track)
  {
  mpeg_header mph;
    
  int frame_bytes;
  int num_samples;
  int new_samples;
  int samples_decoded = 0;
  int bytes_decoded;
  int bytes_used, bytes_skipped;
  int64_t chunk_size;
  quicktime_audio_map_t *track_map = &(file->atracks[track]);
  quicktime_ffmpeg_audio_codec_t *codec = ((quicktime_codec_t*)track_map->codec)->priv;

  /* Read chunk */
  
  chunk_size = lqt_append_audio_chunk(file,
                                      track, track_map->current_chunk,
                                      &(codec->chunk_buffer),
                                      &(codec->chunk_buffer_alloc),
                                      codec->bytes_in_chunk_buffer);

  
  if(!chunk_size)
    {
    /* If the codec is mp3, make sure to decode the very last frame */

    if((codec->avctx->codec_id == CODEC_ID_MP3) &&
       (codec->bytes_in_chunk_buffer >= 4))
      {
      if(!decode_header(&mph, codec->chunk_buffer, (const mpeg_header*)0))
        {
        lqt_log(file, LQT_LOG_ERROR, LOG_DOMAIN, "Decode header failed");
        return 0;
        }
      if(mph.frame_bytes <= codec->bytes_in_chunk_buffer)
        {
        lqt_log(file, LQT_LOG_ERROR, LOG_DOMAIN, "Huh, frame not decoded?");
        return 0;
        }

      if(codec->chunk_buffer_alloc < mph.frame_bytes + FF_INPUT_BUFFER_PADDING_SIZE)
        {
        codec->chunk_buffer_alloc = mph.frame_bytes + FF_INPUT_BUFFER_PADDING_SIZE;
        codec->chunk_buffer = realloc(codec->chunk_buffer, codec->chunk_buffer_alloc);
        }
      memset(codec->chunk_buffer + codec->bytes_in_chunk_buffer, 0,
             mph.frame_bytes - codec->bytes_in_chunk_buffer + FF_INPUT_BUFFER_PADDING_SIZE);
      num_samples = mph.samples_per_frame;
      codec->bytes_in_chunk_buffer = mph.frame_bytes;
      }
    else
      return 0;
    }
  else
    {
    num_samples = quicktime_chunk_samples(track_map->track, track_map->current_chunk);
    track_map->current_chunk++;
    codec->bytes_in_chunk_buffer += chunk_size;
    }
  

  if(!num_samples)
    {
    return 0;
    }
  /*
   *  For AVIs, chunk samples are not always 100% correct.
   *  Furthermore, there can be a complete mp3 frame from the last chunk!
   */