dv.c
Go to the documentation of this file.
1 /*
2  * General DV muxer/demuxer
3  * Copyright (c) 2003 Roman Shaposhnik
4  *
5  * Many thanks to Dan Dennedy <dan@dennedy.org> for providing wealth
6  * of DV technical info.
7  *
8  * Raw DV format
9  * Copyright (c) 2002 Fabrice Bellard
10  *
11  * 50 Mbps (DVCPRO50) and 100 Mbps (DVCPRO HD) support
12  * Copyright (c) 2006 Daniel Maas <dmaas@maasdigital.com>
13  * Funded by BBC Research & Development
14  *
15  * This file is part of Libav.
16  *
17  * Libav is free software; you can redistribute it and/or
18  * modify it under the terms of the GNU Lesser General Public
19  * License as published by the Free Software Foundation; either
20  * version 2.1 of the License, or (at your option) any later version.
21  *
22  * Libav is distributed in the hope that it will be useful,
23  * but WITHOUT ANY WARRANTY; without even the implied warranty of
24  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
25  * Lesser General Public License for more details.
26  *
27  * You should have received a copy of the GNU Lesser General Public
28  * License along with Libav; if not, write to the Free Software
29  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
30  */
31 #include <time.h>
32 #include "avformat.h"
33 #include "internal.h"
34 #include "libavcodec/dv_profile.h"
35 #include "libavcodec/dvdata.h"
37 #include "libavutil/intreadwrite.h"
38 #include "libavutil/mathematics.h"
39 #include "dv.h"
40 
42  const DVprofile* sys; /* Current DV profile. E.g.: 525/60, 625/50 */
47  uint8_t audio_buf[4][8192];
48  int ach;
49  int frames;
50  uint64_t abytes;
51 };
52 
53 static inline uint16_t dv_audio_12to16(uint16_t sample)
54 {
55  uint16_t shift, result;
56 
57  sample = (sample < 0x800) ? sample : sample | 0xf000;
58  shift = (sample & 0xf00) >> 8;
59 
60  if (shift < 0x2 || shift > 0xd) {
61  result = sample;
62  } else if (shift < 0x8) {
63  shift--;
64  result = (sample - (256 * shift)) << shift;
65  } else {
66  shift = 0xe - shift;
67  result = ((sample + ((256 * shift) + 1)) << shift) - 1;
68  }
69 
70  return result;
71 }
72 
73 /*
74  * This is the dumbest implementation of all -- it simply looks at
75  * a fixed offset and if pack isn't there -- fails. We might want
76  * to have a fallback mechanism for complete search of missing packs.
77  */
78 static const uint8_t* dv_extract_pack(uint8_t* frame, enum dv_pack_type t)
79 {
80  int offs;
81 
82  switch (t) {
83  case dv_audio_source:
84  offs = (80*6 + 80*16*3 + 3);
85  break;
86  case dv_audio_control:
87  offs = (80*6 + 80*16*4 + 3);
88  break;
89  case dv_video_control:
90  offs = (80*5 + 48 + 5);
91  break;
92  default:
93  return NULL;
94  }
95 
96  return frame[offs] == t ? &frame[offs] : NULL;
97 }
98 
99 static const int dv_audio_frequency[3] = {
100  48000, 44100, 32000,
101 };
102 
103 /*
104  * There's a couple of assumptions being made here:
105  * 1. By default we silence erroneous (0x8000/16bit 0x800/12bit) audio samples.
106  * We can pass them upwards when libavcodec will be ready to deal with them.
107  * 2. We don't do software emphasis.
108  * 3. Audio is always returned as 16bit linear samples: 12bit nonlinear samples
109  * are converted into 16bit linear ones.
110  */
111 static int dv_extract_audio(uint8_t *frame, uint8_t **ppcm,
112  const DVprofile *sys)
113 {
114  int size, chan, i, j, d, of, smpls, freq, quant, half_ch;
115  uint16_t lc, rc;
116  const uint8_t* as_pack;
117  uint8_t *pcm, ipcm;
118 
119  as_pack = dv_extract_pack(frame, dv_audio_source);
120  if (!as_pack) /* No audio ? */
121  return 0;
122 
123  smpls = as_pack[1] & 0x3f; /* samples in this frame - min. samples */
124  freq = (as_pack[4] >> 3) & 0x07; /* 0 - 48kHz, 1 - 44,1kHz, 2 - 32kHz */
125  quant = as_pack[4] & 0x07; /* 0 - 16bit linear, 1 - 12bit nonlinear */
126 
127  if (quant > 1)
128  return -1; /* unsupported quantization */
129 
130  if (freq >= FF_ARRAY_ELEMS(dv_audio_frequency))
131  return AVERROR_INVALIDDATA;
132 
133  size = (sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */
134  half_ch = sys->difseg_size / 2;
135 
136  /* We work with 720p frames split in half, thus even frames have
137  * channels 0,1 and odd 2,3. */
138  ipcm = (sys->height == 720 && !(frame[1] & 0x0C)) ? 2 : 0;
139 
140  /* for each DIF channel */
141  for (chan = 0; chan < sys->n_difchan; chan++) {
142  /* next stereo channel (50Mbps and 100Mbps only) */
143  pcm = ppcm[ipcm++];
144  if (!pcm)
145  break;
146 
147  /* for each DIF segment */
148  for (i = 0; i < sys->difseg_size; i++) {
149  frame += 6 * 80; /* skip DIF segment header */
150  if (quant == 1 && i == half_ch) {
151  /* next stereo channel (12bit mode only) */
152  pcm = ppcm[ipcm++];
153  if (!pcm)
154  break;
155  }
156 
157  /* for each AV sequence */
158  for (j = 0; j < 9; j++) {
159  for (d = 8; d < 80; d += 2) {
160  if (quant == 0) { /* 16bit quantization */
161  of = sys->audio_shuffle[i][j] + (d - 8) / 2 * sys->audio_stride;
162  if (of*2 >= size)
163  continue;
164 
165  pcm[of*2] = frame[d+1]; // FIXME: maybe we have to admit
166  pcm[of*2+1] = frame[d]; // that DV is a big-endian PCM
167  if (pcm[of*2+1] == 0x80 && pcm[of*2] == 0x00)
168  pcm[of*2+1] = 0;
169  } else { /* 12bit quantization */
170  lc = ((uint16_t)frame[d] << 4) |
171  ((uint16_t)frame[d+2] >> 4);
172  rc = ((uint16_t)frame[d+1] << 4) |
173  ((uint16_t)frame[d+2] & 0x0f);
174  lc = (lc == 0x800 ? 0 : dv_audio_12to16(lc));
175  rc = (rc == 0x800 ? 0 : dv_audio_12to16(rc));
176 
177  of = sys->audio_shuffle[i%half_ch][j] + (d - 8) / 3 * sys->audio_stride;
178  if (of*2 >= size)
179  continue;
180 
181  pcm[of*2] = lc & 0xff; // FIXME: maybe we have to admit
182  pcm[of*2+1] = lc >> 8; // that DV is a big-endian PCM
183  of = sys->audio_shuffle[i%half_ch+half_ch][j] +
184  (d - 8) / 3 * sys->audio_stride;
185  pcm[of*2] = rc & 0xff; // FIXME: maybe we have to admit
186  pcm[of*2+1] = rc >> 8; // that DV is a big-endian PCM
187  ++d;
188  }
189  }
190 
191  frame += 16 * 80; /* 15 Video DIFs + 1 Audio DIF */
192  }
193  }
194  }
195 
196  return size;
197 }
198 
200 {
201  const uint8_t* as_pack;
202  int freq, stype, smpls, quant, i, ach;
203 
204  as_pack = dv_extract_pack(frame, dv_audio_source);
205  if (!as_pack || !c->sys) { /* No audio ? */
206  c->ach = 0;
207  return 0;
208  }
209 
210  smpls = as_pack[1] & 0x3f; /* samples in this frame - min. samples */
211  freq = (as_pack[4] >> 3) & 0x07; /* 0 - 48kHz, 1 - 44,1kHz, 2 - 32kHz */
212  stype = (as_pack[3] & 0x1f); /* 0 - 2CH, 2 - 4CH, 3 - 8CH */
213  quant = as_pack[4] & 0x07; /* 0 - 16bit linear, 1 - 12bit nonlinear */
214 
215  if (freq >= FF_ARRAY_ELEMS(dv_audio_frequency)) {
217  "Unrecognized audio sample rate index (%d)\n", freq);
218  return 0;
219  }
220 
221  if (stype > 3) {
222  av_log(c->fctx, AV_LOG_ERROR, "stype %d is invalid\n", stype);
223  c->ach = 0;
224  return 0;
225  }
226 
227  /* note: ach counts PAIRS of channels (i.e. stereo channels) */
228  ach = ((int[4]){ 1, 0, 2, 4})[stype];
229  if (ach == 1 && quant && freq == 2)
230  ach = 2;
231 
232  /* Dynamic handling of the audio streams in DV */
233  for (i = 0; i < ach; i++) {
234  if (!c->ast[i]) {
235  c->ast[i] = avformat_new_stream(c->fctx, NULL);
236  if (!c->ast[i])
237  break;
238  avpriv_set_pts_info(c->ast[i], 64, 1, 30000);
241 
242  av_init_packet(&c->audio_pkt[i]);
243  c->audio_pkt[i].size = 0;
244  c->audio_pkt[i].data = c->audio_buf[i];
245  c->audio_pkt[i].stream_index = c->ast[i]->index;
247  }
248  c->ast[i]->codec->sample_rate = dv_audio_frequency[freq];
249  c->ast[i]->codec->channels = 2;
251  c->ast[i]->codec->bit_rate = 2 * dv_audio_frequency[freq] * 16;
252  c->ast[i]->start_time = 0;
253  }
254  c->ach = i;
255 
256  return (c->sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */;
257 }
258 
260 {
261  const uint8_t* vsc_pack;
262  AVCodecContext* avctx;
263  int apt, is16_9;
264  int size = 0;
265 
266  if (c->sys) {
267  avctx = c->vst->codec;
268 
270  c->sys->time_base.den);
271  avctx->time_base= c->sys->time_base;
272  if (!avctx->width){
273  avctx->width = c->sys->width;
274  avctx->height = c->sys->height;
275  }
276  avctx->pix_fmt = c->sys->pix_fmt;
277 
278  /* finding out SAR is a little bit messy */
279  vsc_pack = dv_extract_pack(frame, dv_video_control);
280  apt = frame[4] & 0x07;
281  is16_9 = (vsc_pack && ((vsc_pack[2] & 0x07) == 0x02 ||
282  (!apt && (vsc_pack[2] & 0x07) == 0x07)));
283  c->vst->sample_aspect_ratio = c->sys->sar[is16_9];
284  avctx->bit_rate = av_rescale_q(c->sys->frame_size, (AVRational){8,1},
285  c->sys->time_base);
286  size = c->sys->frame_size;
287  }
288  return size;
289 }
290 
291 /*
292  * The following 3 functions constitute our interface to the world
293  */
294 
296 {
297  DVDemuxContext *c;
298 
299  c = av_mallocz(sizeof(DVDemuxContext));
300  if (!c)
301  return NULL;
302 
303  c->vst = avformat_new_stream(s, NULL);
304  if (!c->vst) {
305  av_free(c);
306  return NULL;
307  }
308 
309  c->fctx = s;
312  c->vst->codec->bit_rate = 25000000;
313  c->vst->start_time = 0;
314 
315  return c;
316 }
317 
319 {
320  int size = -1;
321  int i;
322 
323  for (i = 0; i < c->ach; i++) {
324  if (c->ast[i] && c->audio_pkt[i].size) {
325  *pkt = c->audio_pkt[i];
326  c->audio_pkt[i].size = 0;
327  size = pkt->size;
328  break;
329  }
330  }
331 
332  return size;
333 }
334 
336  uint8_t* buf, int buf_size)
337 {
338  int size, i;
339  uint8_t *ppcm[5] = { 0 };
340 
341  if (buf_size < DV_PROFILE_BYTES ||
342  !(c->sys = avpriv_dv_frame_profile(c->sys, buf, buf_size)) ||
343  buf_size < c->sys->frame_size) {
344  return -1; /* Broken frame, or not enough data */
345  }
346 
347  /* Queueing audio packet */
348  /* FIXME: in case of no audio/bad audio we have to do something */
349  size = dv_extract_audio_info(c, buf);
350  for (i = 0; i < c->ach; i++) {
351  c->audio_pkt[i].size = size;
352  c->audio_pkt[i].pts = c->abytes * 30000 * 8 / c->ast[i]->codec->bit_rate;
353  ppcm[i] = c->audio_buf[i];
354  }
355  if (c->ach)
356  dv_extract_audio(buf, ppcm, c->sys);
357 
358  /* We work with 720p frames split in half, thus even frames have
359  * channels 0,1 and odd 2,3. */
360  if (c->sys->height == 720) {
361  if (buf[1] & 0x0C) {
362  c->audio_pkt[2].size = c->audio_pkt[3].size = 0;
363  } else {
364  c->audio_pkt[0].size = c->audio_pkt[1].size = 0;
365  c->abytes += size;
366  }
367  } else {
368  c->abytes += size;
369  }
370 
371  /* Now it's time to return video packet */
372  size = dv_extract_video_info(c, buf);
373  av_init_packet(pkt);
374  pkt->data = buf;
375  pkt->size = size;
376  pkt->flags |= AV_PKT_FLAG_KEY;
377  pkt->stream_index = c->vst->index;
378  pkt->pts = c->frames;
379 
380  c->frames++;
381 
382  return size;
383 }
384 
386  int64_t timestamp, int flags)
387 {
388  // FIXME: sys may be wrong if last dv_read_packet() failed (buffer is junk)
389  const DVprofile* sys = avpriv_dv_codec_profile(c->vst->codec);
390  int64_t offset;
391  int64_t size = avio_size(s->pb) - s->data_offset;
392  int64_t max_offset = ((size-1) / sys->frame_size) * sys->frame_size;
393 
394  offset = sys->frame_size * timestamp;
395 
396  if (size >= 0 && offset > max_offset) offset = max_offset;
397  else if (offset < 0) offset = 0;
398 
399  return offset + s->data_offset;
400 }
401 
402 void ff_dv_offset_reset(DVDemuxContext *c, int64_t frame_offset)
403 {
404  c->frames= frame_offset;
405  if (c->ach)
406  c->abytes= av_rescale_q(c->frames, c->sys->time_base,
407  (AVRational){8, c->ast[0]->codec->bit_rate});
408  c->audio_pkt[0].size = c->audio_pkt[1].size = 0;
409  c->audio_pkt[2].size = c->audio_pkt[3].size = 0;
410 }
411 
412 /************************************************************
413  * Implementation of the easiest DV storage of all -- raw DV.
414  ************************************************************/
415 
416 typedef struct RawDVContext {
419 } RawDVContext;
420 
422 {
423  unsigned state, marker_pos = 0;
424  RawDVContext *c = s->priv_data;
425 
427  if (!c->dv_demux)
428  return -1;
429 
430  state = avio_rb32(s->pb);
431  while ((state & 0xffffff7f) != 0x1f07003f) {
432  if (s->pb->eof_reached) {
433  av_log(s, AV_LOG_ERROR, "Cannot find DV header.\n");
434  return -1;
435  }
436  if (state == 0x003f0700 || state == 0xff3f0700)
437  marker_pos = avio_tell(s->pb);
438  if (state == 0xff3f0701 && avio_tell(s->pb) - marker_pos == 80) {
439  avio_seek(s->pb, -163, SEEK_CUR);
440  state = avio_rb32(s->pb);
441  break;
442  }
443  state = (state << 8) | avio_r8(s->pb);
444  }
445  AV_WB32(c->buf, state);
446 
447  if (avio_read(s->pb, c->buf + 4, DV_PROFILE_BYTES - 4) <= 0 ||
448  avio_seek(s->pb, -DV_PROFILE_BYTES, SEEK_CUR) < 0)
449  return AVERROR(EIO);
450 
452  if (!c->dv_demux->sys) {
453  av_log(s, AV_LOG_ERROR, "Can't determine profile of DV input stream.\n");
454  return -1;
455  }
456 
458  c->dv_demux->sys->time_base);
459 
460  return 0;
461 }
462 
463 
465 {
466  int size;
467  RawDVContext *c = s->priv_data;
468 
469  size = avpriv_dv_get_packet(c->dv_demux, pkt);
470 
471  if (size < 0) {
472  if (!c->dv_demux->sys)
473  return AVERROR(EIO);
474  size = c->dv_demux->sys->frame_size;
475  if (avio_read(s->pb, c->buf, size) <= 0)
476  return AVERROR(EIO);
477 
478  size = avpriv_dv_produce_packet(c->dv_demux, pkt, c->buf, size);
479  }
480 
481  return size;
482 }
483 
484 static int dv_read_seek(AVFormatContext *s, int stream_index,
485  int64_t timestamp, int flags)
486 {
487  RawDVContext *r = s->priv_data;
488  DVDemuxContext *c = r->dv_demux;
489  int64_t offset = dv_frame_offset(s, c, timestamp, flags);
490 
491  if (avio_seek(s->pb, offset, SEEK_SET) < 0)
492  return -1;
493 
494  ff_dv_offset_reset(c, offset / c->sys->frame_size);
495  return 0;
496 }
497 
499 {
500  RawDVContext *c = s->priv_data;
501  av_free(c->dv_demux);
502  return 0;
503 }
504 
505 static int dv_probe(AVProbeData *p)
506 {
507  unsigned state, marker_pos = 0;
508  int i;
509  int matches = 0;
510  int secondary_matches = 0;
511 
512  if (p->buf_size < 5)
513  return 0;
514 
515  state = AV_RB32(p->buf);
516  for (i = 4; i < p->buf_size; i++) {
517  if ((state & 0xffffff7f) == 0x1f07003f)
518  matches++;
519  // any section header, also with seq/chan num != 0,
520  // should appear around every 12000 bytes, at least 10 per frame
521  if ((state & 0xff07ff7f) == 0x1f07003f)
522  secondary_matches++;
523  if (state == 0x003f0700 || state == 0xff3f0700)
524  marker_pos = i;
525  if (state == 0xff3f0701 && i - marker_pos == 80)
526  matches++;
527  state = (state << 8) | p->buf[i];
528  }
529 
530  if (matches && p->buf_size / matches < 1024*1024) {
531  if (matches > 4 || (secondary_matches >= 10 && p->buf_size / secondary_matches < 24000))
532  return AVPROBE_SCORE_MAX*3/4; // not max to avoid dv in mov to match
533  return AVPROBE_SCORE_MAX/4;
534  }
535  return 0;
536 }
537 
538 #if CONFIG_DV_DEMUXER
539 AVInputFormat ff_dv_demuxer = {
540  .name = "dv",
541  .long_name = NULL_IF_CONFIG_SMALL("DV (Digital Video)"),
542  .priv_data_size = sizeof(RawDVContext),
543  .read_probe = dv_probe,
548  .extensions = "dv,dif",
549 };
550 #endif