view libmpdemux/demux_roq.c @ 9660:c2d23e02522b

improvements to detc filter: -> use of 8x8 blocks rather than 16x16 to better localize the search for interlacing. this helps detect interlacing in very small motions, e.g. mouths in anime. -> removed some redundant conditions in the logic -> looser condition for detecting lacing and more forgiving of slight mismatches between fields from the two telecine frames to make up for quantization noise in low quality encodes. this code is still mostly experimental but probably better than the old version, so maybe it should be backported to 0.90...?

/*
        RoQ file demuxer for the MPlayer program
        by Mike Melanson
	based on Dr. Tim Ferguson's RoQ document found at:
	  http://www.csse.monash.edu.au/~timf/videocodec.html
*/

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>

#include "config.h"
#include "mp_msg.h"
#include "help_mp.h"

#include "stream.h"
#include "demuxer.h"
#include "stheader.h"

#define RoQ_INFO           0x1001
#define RoQ_QUAD_CODEBOOK  0x1002
#define RoQ_QUAD_VQ        0x1011
#define RoQ_SOUND_MONO     0x1020
#define RoQ_SOUND_STEREO   0x1021

#define CHUNK_TYPE_AUDIO 0
#define CHUNK_TYPE_VIDEO 1

typedef struct roq_chunk_t
{
  int chunk_type;
  off_t chunk_offset;
  int chunk_size;

  float video_chunk_number;  // in the case of a video chunk
  int running_audio_sample_count;  // for an audio chunk
} roq_chunk_t;

typedef struct roq_data_t
{
  int total_chunks;
  int current_chunk;
  int total_video_chunks;
  int total_audio_sample_count;
  roq_chunk_t *chunks;
} roq_data_t;

// Check if a stream qualifies as a RoQ file based on the magic numbers
// at the start of the file:
//  84 10 FF FF FF FF xx xx
int roq_check_file(demuxer_t *demuxer)
{
  stream_reset(demuxer->stream);
  stream_seek(demuxer->stream, 0);

  if ((stream_read_dword(demuxer->stream) == 0x8410FFFF) &&
      ((stream_read_dword(demuxer->stream) & 0xFFFF0000) == 0xFFFF0000))
    return 1;
  else
    return 0;
}

// return value:
//     0 = EOF or no stream found
//     1 = successfully read a packet
int demux_roq_fill_buffer(demuxer_t *demuxer)
{
  sh_video_t *sh_video = demuxer->video->sh;
  roq_data_t *roq_data = (roq_data_t *)demuxer->priv;
  roq_chunk_t roq_chunk;

  if (roq_data->current_chunk >= roq_data->total_chunks)
    return 0;

  roq_chunk = roq_data->chunks[roq_data->current_chunk];

  // make sure we're at the right place in the stream and fetch the chunk
  stream_seek(demuxer->stream, roq_chunk.chunk_offset);

  if (roq_chunk.chunk_type == CHUNK_TYPE_AUDIO)
    ds_read_packet(demuxer->audio, demuxer->stream, roq_chunk.chunk_size,
      0,
      roq_chunk.chunk_offset, 0);
  else
    ds_read_packet(demuxer->video, demuxer->stream, roq_chunk.chunk_size,
      roq_chunk.video_chunk_number / sh_video->fps,
      roq_chunk.chunk_offset, 0);

  roq_data->current_chunk++;
  return 1;
}

demuxer_t* demux_open_roq(demuxer_t* demuxer)
{
  sh_video_t *sh_video = NULL;
  sh_audio_t *sh_audio = NULL;

  roq_data_t *roq_data = (roq_data_t *)malloc(sizeof(roq_data_t));
  int chunk_id;
  int chunk_size;
  int chunk_arg;
  int last_chunk_id = 0;
  int largest_audio_chunk = 0;
  int fps;

  roq_data->total_chunks = 0;
  roq_data->current_chunk = 0;
  roq_data->total_video_chunks = 0;
  roq_data->chunks = NULL;

  // position the stream and start traversing
  stream_seek(demuxer->stream, 6);
  fps = stream_read_word_le(demuxer->stream);
  while (!stream_eof(demuxer->stream))
  {
    chunk_id = stream_read_word_le(demuxer->stream);
    chunk_size = stream_read_dword_le(demuxer->stream);
    chunk_arg = stream_read_word_le(demuxer->stream);

    // this is the only useful header info in the file
    if (chunk_id == RoQ_INFO)
    {
      // there should only be one RoQ_INFO chunk per file
      if (sh_video)
      {
        mp_msg(MSGT_DECVIDEO, MSGL_WARN, "Found more than one RoQ_INFO chunk\n");
        stream_skip(demuxer->stream, 8);
      }
      else
      {
        // this is a good opportunity to create a video stream header
        sh_video = new_sh_video(demuxer, 0);
        // make sure the demuxer knows about the new stream header
        demuxer->video->sh = sh_video;
        // make sure that the video demuxer stream header knows about its
        // parent video demuxer stream
        sh_video->ds = demuxer->video;

        sh_video->disp_w = stream_read_word_le(demuxer->stream);
        sh_video->disp_h = stream_read_word_le(demuxer->stream);
        stream_skip(demuxer->stream, 4);

        // custom fourcc for internal MPlayer use
        sh_video->format = mmioFOURCC('R', 'o', 'Q', 'V');

        // constant frame rate
        sh_video->fps = fps;
        sh_video->frametime = 1 / sh_video->fps;
      }
    }
    else if ((chunk_id == RoQ_SOUND_MONO) ||
      (chunk_id == RoQ_SOUND_STEREO))
    {
      // create the audio stream header if it hasn't been created it
      if (sh_audio == NULL)
      {
        // make the header first
        sh_audio = new_sh_audio(demuxer, 0);
        // make sure the demuxer knows about the new stream header
        demuxer->audio->sh = sh_audio;
        // make sure that the audio demuxer stream header knows about its
        // parent audio demuxer stream
        sh_audio->ds = demuxer->audio;

        // go through the bother of making a WAVEFORMATEX structure
        sh_audio->wf = (WAVEFORMATEX *)malloc(sizeof(WAVEFORMATEX));

        // custom fourcc for internal MPlayer use
        sh_audio->format = mmioFOURCC('R', 'o', 'Q', 'A');
        if (chunk_id == RoQ_SOUND_STEREO)
          sh_audio->wf->nChannels = 2;
        else
          sh_audio->wf->nChannels = 1;
        // always 22KHz, 16-bit
        sh_audio->wf->nSamplesPerSec = 22050;
        sh_audio->wf->wBitsPerSample = 16;
      }

      // index the chunk
      roq_data->chunks = (roq_chunk_t *)realloc(roq_data->chunks,
        (roq_data->total_chunks + 1) * sizeof (roq_chunk_t));
      roq_data->chunks[roq_data->total_chunks].chunk_type = CHUNK_TYPE_AUDIO;
      roq_data->chunks[roq_data->total_chunks].chunk_offset = 
        stream_tell(demuxer->stream) - 8;
      roq_data->chunks[roq_data->total_chunks].chunk_size = chunk_size + 8;
      roq_data->chunks[roq_data->total_chunks].running_audio_sample_count =
        roq_data->total_audio_sample_count;

      // audio housekeeping
      if (chunk_size > largest_audio_chunk)
        largest_audio_chunk = chunk_size;
      roq_data->total_audio_sample_count += 
        (chunk_size / sh_audio->wf->nChannels);

      stream_skip(demuxer->stream, chunk_size);
      roq_data->total_chunks++;
    }
    else if ((chunk_id == RoQ_QUAD_CODEBOOK) ||
      ((chunk_id == RoQ_QUAD_VQ) && (last_chunk_id != RoQ_QUAD_CODEBOOK)))
    {
      // index a new chunk if it's a codebook or quad VQ not following a
      // codebook
      roq_data->chunks = (roq_chunk_t *)realloc(roq_data->chunks,
        (roq_data->total_chunks + 1) * sizeof (roq_chunk_t));
      roq_data->chunks[roq_data->total_chunks].chunk_type = CHUNK_TYPE_VIDEO;
      roq_data->chunks[roq_data->total_chunks].chunk_offset = 
        stream_tell(demuxer->stream) - 8;
      roq_data->chunks[roq_data->total_chunks].chunk_size = chunk_size + 8;
      roq_data->chunks[roq_data->total_chunks].video_chunk_number = 
        roq_data->total_video_chunks++;

      stream_skip(demuxer->stream, chunk_size);
      roq_data->total_chunks++;
    }
    else if ((chunk_id == RoQ_QUAD_VQ) && (last_chunk_id == RoQ_QUAD_CODEBOOK))
    {
      // if it's a quad VQ chunk following a codebook chunk, extend the last
      // chunk
      roq_data->chunks[roq_data->total_chunks - 1].chunk_size += (chunk_size + 8);
      stream_skip(demuxer->stream, chunk_size);
    }
    else if (!stream_eof(demuxer->stream))
    {
        mp_msg(MSGT_DECVIDEO, MSGL_WARN, "Unknown RoQ chunk ID: %04X\n", chunk_id);
    }

    last_chunk_id = chunk_id;
  }

  // minimum output buffer size = largest audio chunk * 2, since each byte
  // in the DPCM encoding effectively represents 1 16-bit sample
  // (store it in wf->nBlockAlign for the time being since init_audio() will
  // step on it anyway)
  if (sh_audio)
    sh_audio->wf->nBlockAlign = largest_audio_chunk * 2;

  roq_data->current_chunk = 0;

  demuxer->priv = roq_data;

  stream_reset(demuxer->stream);

  return demuxer;
}

void demux_close_roq(demuxer_t* demuxer) {
  roq_data_t *roq_data = demuxer->priv;

  if(!roq_data)
    return;
  free(roq_data);
}