FFmpeg
h264_parser.c
Go to the documentation of this file.
1 /*
2  * H.26L/H.264/AVC/JVT/14496-10/... parser
3  * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 /**
23  * @file
24  * H.264 / AVC / MPEG-4 part10 parser.
25  * @author Michael Niedermayer <michaelni@gmx.at>
26  */
27 
28 #define UNCHECKED_BITSTREAM_READER 1
29 
30 #include <stdint.h>
31 
32 #include "libavutil/attributes.h"
33 #include "libavutil/avutil.h"
34 #include "libavutil/error.h"
35 #include "libavutil/log.h"
36 #include "libavutil/mem.h"
37 #include "libavutil/pixfmt.h"
38 
39 #include "avcodec.h"
40 #include "get_bits.h"
41 #include "golomb.h"
42 #include "h264.h"
43 #include "h264dsp.h"
44 #include "h264_parse.h"
45 #include "h264_sei.h"
46 #include "h264_ps.h"
47 #include "h2645_parse.h"
48 #include "h264data.h"
49 #include "mpegutils.h"
50 #include "parser.h"
51 #include "libavutil/refstruct.h"
52 #include "parser_internal.h"
53 #include "startcode.h"
54 
55 typedef struct H264ParseContext {
61  int is_avc;
63  int got_first;
65  uint8_t parse_history[6];
71 
72 static int find_start_code(const uint8_t *buf, int buf_size,
73  int buf_index, int next_avc)
74 {
75  uint32_t state = -1;
76 
77  buf_index = avpriv_find_start_code(buf + buf_index, buf + next_avc + 1, &state) - buf - 1;
78 
79  return FFMIN(buf_index, buf_size);
80 }
81 
82 static int h264_find_frame_end(H264ParseContext *p, const uint8_t *buf,
83  int buf_size, void *logctx)
84 {
85  int i, j;
86  uint32_t state;
87  ParseContext *pc = &p->pc;
88 
89  int next_avc = p->is_avc ? 0 : buf_size;
90 // mb_addr= pc->mb_addr - 1;
91  state = pc->state;
92  if (state > 13)
93  state = 7;
94 
95  if (p->is_avc && !p->nal_length_size)
96  av_log(logctx, AV_LOG_ERROR, "AVC-parser: nal length size invalid\n");
97 
98  for (i = 0; i < buf_size; i++) {
99  if (i >= next_avc) {
100  int64_t nalsize = 0;
101  i = next_avc;
102  for (j = 0; j < p->nal_length_size; j++)
103  nalsize = (nalsize << 8) | buf[i++];
104  if (!nalsize || nalsize > buf_size - i) {
105  av_log(logctx, AV_LOG_ERROR, "AVC-parser: nal size %"PRId64" "
106  "remaining %d\n", nalsize, buf_size - i);
107  return buf_size;
108  }
109  next_avc = i + nalsize;
110  state = 5;
111  }
112 
113  if (state == 7) {
114  i += p->h264dsp.startcode_find_candidate(buf + i, next_avc - i);
115  if (i < next_avc)
116  state = 2;
117  } else if (state <= 2) {
118  if (buf[i] == 1)
119  state ^= 5; // 2->7, 1->4, 0->5
120  else if (buf[i])
121  state = 7;
122  else
123  state >>= 1; // 2->1, 1->0, 0->0
124  } else if (state <= 5) {
125  int nalu_type = buf[i] & 0x1F;
126  if (nalu_type == H264_NAL_SEI || nalu_type == H264_NAL_SPS ||
127  nalu_type == H264_NAL_PPS || nalu_type == H264_NAL_AUD) {
128  if (pc->frame_start_found) {
129  i++;
130  goto found;
131  }
132  } else if (nalu_type == H264_NAL_SLICE || nalu_type == H264_NAL_DPA ||
133  nalu_type == H264_NAL_IDR_SLICE) {
134  state += 8;
135  continue;
136  }
137  state = 7;
138  } else {
139  unsigned int mb, last_mb = p->parse_last_mb;
140  GetBitContext gb;
141  p->parse_history[p->parse_history_count++] = buf[i];
142 
143  init_get_bits(&gb, p->parse_history, 8*p->parse_history_count);
144  mb= get_ue_golomb_long(&gb);
145  if (get_bits_left(&gb) > 0 || p->parse_history_count > 5) {
146  p->parse_last_mb = mb;
147  if (pc->frame_start_found) {
148  if (mb <= last_mb) {
149  i -= p->parse_history_count - 1;
150  p->parse_history_count = 0;
151  goto found;
152  }
153  } else
154  pc->frame_start_found = 1;
155  p->parse_history_count = 0;
156  state = 7;
157  }
158  }
159  }
160  pc->state = state;
161  if (p->is_avc)
162  return next_avc;
163  return END_NOT_FOUND;
164 
165 found:
166  pc->state = 7;
167  pc->frame_start_found = 0;
168  if (p->is_avc)
169  return next_avc;
170  return i - (state & 5);
171 }
172 
174  void *logctx)
175 {
177  int slice_type_nos = s->pict_type & 3;
178  H264ParseContext *p = s->priv_data;
179  int list_count, ref_count[2];
180 
181 
182  if (p->ps.pps->redundant_pic_cnt_present)
183  get_ue_golomb(gb); // redundant_pic_count
184 
185  if (slice_type_nos == AV_PICTURE_TYPE_B)
186  get_bits1(gb); // direct_spatial_mv_pred
187 
188  if (ff_h264_parse_ref_count(&list_count, ref_count, gb, p->ps.pps,
189  slice_type_nos, p->picture_structure, logctx) < 0)
190  return AVERROR_INVALIDDATA;
191 
192  if (slice_type_nos != AV_PICTURE_TYPE_I) {
193  int list;
194  for (list = 0; list < list_count; list++) {
195  if (get_bits1(gb)) {
196  int index;
197  for (index = 0; ; index++) {
198  unsigned int reordering_of_pic_nums_idc = get_ue_golomb_31(gb);
199 
200  if (reordering_of_pic_nums_idc < 3)
201  get_ue_golomb_long(gb);
202  else if (reordering_of_pic_nums_idc > 3) {
203  av_log(logctx, AV_LOG_ERROR,
204  "illegal reordering_of_pic_nums_idc %d\n",
205  reordering_of_pic_nums_idc);
206  return AVERROR_INVALIDDATA;
207  } else
208  break;
209 
210  if (index >= ref_count[list]) {
211  av_log(logctx, AV_LOG_ERROR,
212  "reference count %d overflow\n", index);
213  return AVERROR_INVALIDDATA;
214  }
215  }
216  }
217  }
218  }
219 
220  if ((p->ps.pps->weighted_pred && slice_type_nos == AV_PICTURE_TYPE_P) ||
221  (p->ps.pps->weighted_bipred_idc == 1 && slice_type_nos == AV_PICTURE_TYPE_B))
222  ff_h264_pred_weight_table(gb, p->ps.sps, ref_count, slice_type_nos,
223  &pwt, p->picture_structure, logctx);
224 
225  if (get_bits1(gb)) { // adaptive_ref_pic_marking_mode_flag
226  int i;
227  for (i = 0; i < H264_MAX_MMCO_COUNT; i++) {
228  if (get_bits_left(gb) < 1)
229  return AVERROR_INVALIDDATA;
230 
231  MMCOOpcode opcode = get_ue_golomb_31(gb);
232  if (opcode > (unsigned) MMCO_LONG) {
233  av_log(logctx, AV_LOG_ERROR,
234  "illegal memory management control operation %d\n",
235  opcode);
236  return AVERROR_INVALIDDATA;
237  }
238  if (opcode == MMCO_END)
239  return 0;
240  else if (opcode == MMCO_RESET)
241  return 1;
242 
243  if (opcode == MMCO_SHORT2UNUSED || opcode == MMCO_SHORT2LONG)
244  get_ue_golomb_long(gb); // difference_of_pic_nums_minus1
245  if (opcode == MMCO_SHORT2LONG || opcode == MMCO_LONG2UNUSED ||
246  opcode == MMCO_LONG || opcode == MMCO_SET_MAX_LONG)
247  get_ue_golomb_31(gb);
248  }
249  }
250 
251  return 0;
252 }
253 
254 /**
255  * Parse NAL units of found picture and decode some basic information.
256  *
257  * @param s parser context.
258  * @param avctx codec context.
259  * @param buf buffer with field/frame data.
260  * @param buf_size size of the buffer.
261  */
263  AVCodecContext *avctx,
264  const uint8_t * const buf, int buf_size)
265 {
266  H264ParseContext *p = s->priv_data;
267  H2645RBSP rbsp = { NULL };
268  H2645NAL nal = { NULL };
269  int buf_index, next_avc;
270  unsigned int pps_id;
271  unsigned int slice_type;
272  int state = -1, got_reset = 0;
273  int q264 = buf_size >=4 && !memcmp("Q264", buf, 4);
274  int field_poc[2];
275  int ret;
276 
277  /* set some sane default values */
278  s->pict_type = AV_PICTURE_TYPE_I;
279  s->key_frame = 0;
280  s->picture_structure = AV_PICTURE_STRUCTURE_UNKNOWN;
281 
282  ff_h264_sei_uninit(&p->sei);
283  p->sei.common.frame_packing.arrangement_cancel_flag = -1;
284  p->sei.common.unregistered.x264_build = -1;
285 
286  if (!buf_size)
287  return 0;
288 
290  if (!rbsp.rbsp_buffer)
291  return AVERROR(ENOMEM);
292 
293  buf_index = 0;
294  next_avc = p->is_avc ? 0 : buf_size;
295  for (;;) {
296  const SPS *sps;
297  int src_length, consumed, nalsize = 0;
298 
299  if (buf_index >= next_avc) {
300  nalsize = get_nalsize(p->nal_length_size, buf, buf_size, &buf_index, avctx);
301  if (nalsize < 0)
302  break;
303  next_avc = buf_index + nalsize;
304  } else {
305  buf_index = find_start_code(buf, buf_size, buf_index, next_avc);
306  if (buf_index >= buf_size)
307  break;
308  if (buf_index >= next_avc)
309  continue;
310  }
311  src_length = next_avc - buf_index;
312 
313  state = buf[buf_index];
314  switch (state & 0x1f) {
315  case H264_NAL_SLICE:
316  case H264_NAL_IDR_SLICE:
317  // Do not walk the whole buffer just to decode slice header
318  if ((state & 0x1f) == H264_NAL_IDR_SLICE || ((state >> 5) & 0x3) == 0) {
319  /* IDR or disposable slice
320  * No need to decode many bytes because MMCOs shall not be present. */
321  if (src_length > 60)
322  src_length = 60;
323  } else {
324  /* To decode up to MMCOs */
325  if (src_length > 1000)
326  src_length = 1000;
327  }
328  break;
329  }
330  consumed = ff_h2645_extract_rbsp(buf + buf_index, src_length, &rbsp, &nal, 1);
331  if (consumed < 0)
332  break;
333 
334  buf_index += consumed;
335 
336  ret = init_get_bits8(&nal.gb, nal.data, nal.size);
337  if (ret < 0)
338  goto fail;
339  get_bits1(&nal.gb);
340  nal.ref_idc = get_bits(&nal.gb, 2);
341  nal.type = get_bits(&nal.gb, 5);
342 
343  switch (nal.type) {
344  case H264_NAL_SPS:
345  ff_h264_decode_seq_parameter_set(&nal.gb, avctx, &p->ps, 0);
346  break;
347  case H264_NAL_PPS:
348  ff_h264_decode_picture_parameter_set(&nal.gb, avctx, &p->ps,
349  nal.size_bits);
350  break;
351  case H264_NAL_SEI:
352  ff_h264_sei_decode(&p->sei, &nal.gb, &p->ps, avctx);
353  break;
354  case H264_NAL_IDR_SLICE:
355  s->key_frame = 1;
356 
357  p->poc.prev_frame_num = 0;
358  p->poc.prev_frame_num_offset = 0;
359  p->poc.prev_poc_msb =
360  p->poc.prev_poc_lsb = 0;
361  /* fall through */
362  case H264_NAL_SLICE:
363  get_ue_golomb_long(&nal.gb); // skip first_mb_in_slice
364  slice_type = get_ue_golomb_31(&nal.gb);
365  s->pict_type = ff_h264_golomb_to_pict_type[slice_type % 5];
366  if (p->sei.recovery_point.recovery_frame_cnt >= 0) {
367  /* key frame, since recovery_frame_cnt is set */
368  s->key_frame = 1;
369  }
370  pps_id = get_ue_golomb(&nal.gb);
371  if (pps_id >= MAX_PPS_COUNT) {
372  av_log(avctx, AV_LOG_ERROR,
373  "pps_id %u out of range\n", pps_id);
374  goto fail;
375  }
376  if (!p->ps.pps_list[pps_id]) {
377  av_log(avctx, AV_LOG_ERROR,
378  "non-existing PPS %u referenced\n", pps_id);
379  goto fail;
380  }
381 
382  av_refstruct_replace(&p->ps.pps, p->ps.pps_list[pps_id]);
383  p->ps.sps = p->ps.pps->sps;
384  sps = p->ps.sps;
385 
386  // heuristic to detect non marked keyframes
387  if (p->ps.sps->ref_frame_count <= 1 && p->ps.pps->ref_count[0] <= 1 && s->pict_type == AV_PICTURE_TYPE_I)
388  s->key_frame = 1;
389 
390  p->poc.frame_num = get_bits(&nal.gb, sps->log2_max_frame_num);
391 
392  s->coded_width = 16 * sps->mb_width;
393  s->coded_height = 16 * sps->mb_height;
394  s->width = s->coded_width - (sps->crop_right + sps->crop_left);
395  s->height = s->coded_height - (sps->crop_top + sps->crop_bottom);
396  if (s->width <= 0 || s->height <= 0) {
397  s->width = s->coded_width;
398  s->height = s->coded_height;
399  }
400 
401  switch (sps->bit_depth_luma) {
402  case 9:
403  if (sps->chroma_format_idc == 3) s->format = AV_PIX_FMT_YUV444P9;
404  else if (sps->chroma_format_idc == 2) s->format = AV_PIX_FMT_YUV422P9;
405  else s->format = AV_PIX_FMT_YUV420P9;
406  break;
407  case 10:
408  if (sps->chroma_format_idc == 3) s->format = AV_PIX_FMT_YUV444P10;
409  else if (sps->chroma_format_idc == 2) s->format = AV_PIX_FMT_YUV422P10;
410  else s->format = AV_PIX_FMT_YUV420P10;
411  break;
412  case 8:
413  if (sps->chroma_format_idc == 3) s->format = AV_PIX_FMT_YUV444P;
414  else if (sps->chroma_format_idc == 2) s->format = AV_PIX_FMT_YUV422P;
415  else s->format = AV_PIX_FMT_YUV420P;
416  break;
417  default:
418  s->format = AV_PIX_FMT_NONE;
419  }
420 
421  avctx->profile = ff_h264_get_profile(sps);
422  avctx->level = sps->level_idc;
423 
424  if (sps->frame_mbs_only_flag) {
425  p->picture_structure = PICT_FRAME;
426  } else {
427  if (get_bits1(&nal.gb)) { // field_pic_flag
428  p->picture_structure = PICT_TOP_FIELD + get_bits1(&nal.gb); // bottom_field_flag
429  } else {
430  p->picture_structure = PICT_FRAME;
431  }
432  }
433 
434  if (nal.type == H264_NAL_IDR_SLICE)
435  get_ue_golomb_long(&nal.gb); /* idr_pic_id */
436  if (sps->poc_type == 0) {
437  p->poc.poc_lsb = get_bits(&nal.gb, sps->log2_max_poc_lsb);
438 
439  if (p->ps.pps->pic_order_present == 1 &&
440  p->picture_structure == PICT_FRAME)
441  p->poc.delta_poc_bottom = get_se_golomb(&nal.gb);
442  }
443 
444  if (sps->poc_type == 1 &&
445  !sps->delta_pic_order_always_zero_flag) {
446  p->poc.delta_poc[0] = get_se_golomb(&nal.gb);
447 
448  if (p->ps.pps->pic_order_present == 1 &&
449  p->picture_structure == PICT_FRAME)
450  p->poc.delta_poc[1] = get_se_golomb(&nal.gb);
451  }
452 
453  /* Decode POC of this picture.
454  * The prev_ values needed for decoding POC of the next picture are not set here. */
455  field_poc[0] = field_poc[1] = INT_MAX;
456  ret = ff_h264_init_poc(field_poc, &s->output_picture_number, sps,
457  &p->poc, p->picture_structure, nal.ref_idc);
458  if (ret < 0)
459  goto fail;
460 
461  /* Continue parsing to check if MMCO_RESET is present.
462  * FIXME: MMCO_RESET could appear in non-first slice.
463  * Maybe, we should parse all undisposable non-IDR slice of this
464  * picture until encountering MMCO_RESET in a slice of it. */
465  if (nal.ref_idc && nal.type != H264_NAL_IDR_SLICE) {
466  got_reset = scan_mmco_reset(s, &nal.gb, avctx);
467  if (got_reset < 0)
468  goto fail;
469  }
470 
471  /* Set up the prev_ values for decoding POC of the next picture. */
472  p->poc.prev_frame_num = got_reset ? 0 : p->poc.frame_num;
473  p->poc.prev_frame_num_offset = got_reset ? 0 : p->poc.frame_num_offset;
474  if (nal.ref_idc != 0) {
475  if (!got_reset) {
476  p->poc.prev_poc_msb = p->poc.poc_msb;
477  p->poc.prev_poc_lsb = p->poc.poc_lsb;
478  } else {
479  p->poc.prev_poc_msb = 0;
480  p->poc.prev_poc_lsb =
481  p->picture_structure == PICT_BOTTOM_FIELD ? 0 : field_poc[0];
482  }
483  }
484 
485  if (p->sei.picture_timing.present) {
486  ret = ff_h264_sei_process_picture_timing(&p->sei.picture_timing,
487  sps, avctx);
488  if (ret < 0) {
489  av_log(avctx, AV_LOG_ERROR, "Error processing the picture timing SEI\n");
490  p->sei.picture_timing.present = 0;
491  }
492  }
493 
494  if (sps->pic_struct_present_flag && p->sei.picture_timing.present) {
495  switch (p->sei.picture_timing.pic_struct) {
498  s->repeat_pict = 0;
499  break;
503  s->repeat_pict = 1;
504  break;
507  s->repeat_pict = 2;
508  break;
510  s->repeat_pict = 3;
511  break;
513  s->repeat_pict = 5;
514  break;
515  default:
516  s->repeat_pict = p->picture_structure == PICT_FRAME ? 1 : 0;
517  break;
518  }
519  } else {
520  s->repeat_pict = p->picture_structure == PICT_FRAME ? 1 : 0;
521  }
522 
523  if (p->picture_structure == PICT_FRAME) {
524  s->picture_structure = AV_PICTURE_STRUCTURE_FRAME;
525  if (sps->pic_struct_present_flag && p->sei.picture_timing.present) {
526  switch (p->sei.picture_timing.pic_struct) {
529  s->field_order = AV_FIELD_TT;
530  break;
533  s->field_order = AV_FIELD_BB;
534  break;
535  default:
536  s->field_order = AV_FIELD_PROGRESSIVE;
537  break;
538  }
539  } else {
540  if (field_poc[0] < field_poc[1])
541  s->field_order = AV_FIELD_TT;
542  else if (field_poc[0] > field_poc[1])
543  s->field_order = AV_FIELD_BB;
544  else
545  s->field_order = AV_FIELD_PROGRESSIVE;
546  }
547  } else {
548  if (p->picture_structure == PICT_TOP_FIELD)
549  s->picture_structure = AV_PICTURE_STRUCTURE_TOP_FIELD;
550  else
551  s->picture_structure = AV_PICTURE_STRUCTURE_BOTTOM_FIELD;
552  if (p->poc.frame_num == p->last_frame_num &&
553  p->last_picture_structure != AV_PICTURE_STRUCTURE_UNKNOWN &&
554  p->last_picture_structure != AV_PICTURE_STRUCTURE_FRAME &&
555  p->last_picture_structure != s->picture_structure) {
556  if (p->last_picture_structure == AV_PICTURE_STRUCTURE_TOP_FIELD)
557  s->field_order = AV_FIELD_TT;
558  else
559  s->field_order = AV_FIELD_BB;
560  } else {
561  s->field_order = AV_FIELD_UNKNOWN;
562  }
563  p->last_picture_structure = s->picture_structure;
564  p->last_frame_num = p->poc.frame_num;
565  }
566  if (sps->timing_info_present_flag) {
567  int64_t den = sps->time_scale;
568  if (p->sei.common.unregistered.x264_build < 44U)
569  den *= 2;
570  av_reduce(&avctx->framerate.den, &avctx->framerate.num,
571  sps->num_units_in_tick * 2, den, 1 << 30);
572  }
573 
574  av_freep(&rbsp.rbsp_buffer);
575  return 0; /* no need to evaluate the rest */
576  }
577  }
578  if (q264) {
579  av_freep(&rbsp.rbsp_buffer);
580  return 0;
581  }
582  /* didn't find a picture! */
583  av_log(avctx, AV_LOG_ERROR, "missing picture in access unit with size %d\n", buf_size);
584 fail:
585  av_freep(&rbsp.rbsp_buffer);
586  return -1;
587 }
588 
590  AVCodecContext *avctx,
591  const uint8_t **poutbuf, int *poutbuf_size,
592  const uint8_t *buf, int buf_size)
593 {
594  H264ParseContext *p = s->priv_data;
595  ParseContext *pc = &p->pc;
596  AVRational time_base = { 0, 1 };
597  int next;
598 
599  if (!p->got_first) {
600  p->got_first = 1;
601  if (avctx->extradata_size) {
603  &p->ps, &p->is_avc, &p->nal_length_size,
604  avctx->err_recognition, avctx);
605  }
606  }
607 
608  if (s->flags & PARSER_FLAG_COMPLETE_FRAMES) {
609  next = buf_size;
610  } else {
611  next = h264_find_frame_end(p, buf, buf_size, avctx);
612 
613  if (ff_combine_frame(pc, next, &buf, &buf_size) < 0) {
614  *poutbuf = NULL;
615  *poutbuf_size = 0;
616  return buf_size;
617  }
618 
619  if (next < 0 && next != END_NOT_FOUND) {
620  av_assert1(pc->last_index + next >= 0);
621  h264_find_frame_end(p, &pc->buffer[pc->last_index + next], -next, avctx); // update state
622  }
623  }
624 
625  parse_nal_units(s, avctx, buf, buf_size);
626 
627  if (avctx->framerate.num)
628  time_base = av_inv_q(av_mul_q(avctx->framerate, (AVRational){2, 1}));
629  if (p->sei.picture_timing.cpb_removal_delay >= 0) {
630  s->dts_sync_point = p->sei.buffering_period.present;
631  s->dts_ref_dts_delta = p->sei.picture_timing.cpb_removal_delay;
632  s->pts_dts_delta = p->sei.picture_timing.dpb_output_delay;
633  } else {
634  s->dts_sync_point = INT_MIN;
635  s->dts_ref_dts_delta = INT_MIN;
636  s->pts_dts_delta = INT_MIN;
637  }
638 
639  if (s->flags & PARSER_FLAG_ONCE) {
640  s->flags &= PARSER_FLAG_COMPLETE_FRAMES;
641  }
642 
643  if (s->dts_sync_point >= 0) {
644  int64_t den = time_base.den * (int64_t)avctx->pkt_timebase.num;
645  if (den > 0) {
646  int64_t num = time_base.num * (int64_t)avctx->pkt_timebase.den;
647  if (s->dts != AV_NOPTS_VALUE) {
648  // got DTS from the stream, update reference timestamp
649  p->reference_dts = av_sat_sub64(s->dts, av_rescale(s->dts_ref_dts_delta, num, den));
650  } else if (p->reference_dts != AV_NOPTS_VALUE) {
651  // compute DTS based on reference timestamp
652  s->dts = av_sat_add64(p->reference_dts, av_rescale(s->dts_ref_dts_delta, num, den));
653  }
654 
655  if (p->reference_dts != AV_NOPTS_VALUE && s->pts == AV_NOPTS_VALUE)
656  s->pts = s->dts + av_rescale(s->pts_dts_delta, num, den);
657 
658  if (s->dts_sync_point > 0)
659  p->reference_dts = s->dts; // new reference
660  }
661  }
662 
663  *poutbuf = buf;
664  *poutbuf_size = buf_size;
665  return next;
666 }
667 
669 {
670  H264ParseContext *p = s->priv_data;
671  ParseContext *pc = &p->pc;
672 
673  av_freep(&pc->buffer);
674 
675  ff_h264_sei_uninit(&p->sei);
676  ff_h264_ps_uninit(&p->ps);
677 }
678 
680 {
681  H264ParseContext *p = s->priv_data;
682 
683  p->reference_dts = AV_NOPTS_VALUE;
684  p->last_frame_num = INT_MAX;
685  ff_h264dsp_init(&p->h264dsp, 8, 1);
686  return 0;
687 }
688 
691  .priv_data_size = sizeof(H264ParseContext),
692  .init = init,
693  .parse = h264_parse,
694  .close = h264_close,
695 };
MMCO_LONG2UNUSED
@ MMCO_LONG2UNUSED
Definition: h264_parse.h:62
H264_NAL_PPS
@ H264_NAL_PPS
Definition: h264.h:42
PICT_FRAME
#define PICT_FRAME
Definition: mpegutils.h:33
H264ParseContext
Definition: h264_parser.c:55
MMCO_LONG
@ MMCO_LONG
Definition: h264_parse.h:66
ff_h264_sei_uninit
void ff_h264_sei_uninit(H264SEIContext *h)
Reset SEI values at the beginning of the frame.
Definition: h264_sei.c:48
h2645_parse.h
get_bits_left
static int get_bits_left(GetBitContext *gb)
Definition: get_bits.h:694
AVERROR
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
H264ParseContext::sei
H264SEIContext sei
Definition: h264_parser.c:60
ff_h264_ps_uninit
void ff_h264_ps_uninit(H264ParamSets *ps)
Uninit H264 param sets structure.
Definition: h264_ps.c:270
h264_close
static av_cold void h264_close(AVCodecParserContext *s)
Definition: h264_parser.c:668
H264ParseContext::parse_last_mb
int parse_last_mb
Definition: h264_parser.c:67
AV_FIELD_PROGRESSIVE
@ AV_FIELD_PROGRESSIVE
Definition: defs.h:213
AV_PICTURE_STRUCTURE_UNKNOWN
@ AV_PICTURE_STRUCTURE_UNKNOWN
unknown
Definition: avcodec.h:2583
AVCodecContext::err_recognition
int err_recognition
Error recognition; may misdetect some more or less valid parts as errors.
Definition: avcodec.h:1406
int64_t
long long int64_t
Definition: coverity.c:34
H264ParseContext::is_avc
int is_avc
Definition: h264_parser.c:61
H264_SEI_PIC_STRUCT_TOP_BOTTOM
@ H264_SEI_PIC_STRUCT_TOP_BOTTOM
3: top field, bottom field, in that order
Definition: h264_sei.h:35
H2645NAL::ref_idc
int ref_idc
H.264 only, nal_ref_idc.
Definition: h2645_parse.h:57
scan_mmco_reset
static int scan_mmco_reset(AVCodecParserContext *s, GetBitContext *gb, void *logctx)
Definition: h264_parser.c:173
h264_parse.h
parser_internal.h
get_ue_golomb
static int get_ue_golomb(GetBitContext *gb)
Read an unsigned Exp-Golomb code in the range 0 to 8190.
Definition: golomb.h:53
H264ParseContext::parse_history
uint8_t parse_history[6]
Definition: h264_parser.c:65
AV_PIX_FMT_YUV420P10
#define AV_PIX_FMT_YUV420P10
Definition: pixfmt.h:539
PICT_BOTTOM_FIELD
#define PICT_BOTTOM_FIELD
Definition: mpegutils.h:32
ff_h264_pred_weight_table
int ff_h264_pred_weight_table(GetBitContext *gb, const SPS *sps, const int *ref_count, int slice_type_nos, H264PredWeightTable *pwt, int picture_structure, void *logctx)
Definition: h264_parse.c:30
H264ParseContext::picture_structure
int picture_structure
Definition: h264_parser.c:64
mpegutils.h
MMCOOpcode
MMCOOpcode
Memory management control operation opcode.
Definition: h264_parse.h:59
ParseContext::state
uint32_t state
contains the last few bytes in MSB order
Definition: parser.h:33
init_get_bits
static int init_get_bits(GetBitContext *s, const uint8_t *buffer, int bit_size)
Initialize GetBitContext.
Definition: get_bits.h:517
ParseContext::last_index
int last_index
Definition: parser.h:31
H264ParseContext::got_first
int got_first
Definition: h264_parser.c:63
H2645NAL::size_bits
int size_bits
Size, in bits, of just the data, excluding the stop bit and any trailing padding.
Definition: h2645_parse.h:42
ff_h264_decode_picture_parameter_set
int ff_h264_decode_picture_parameter_set(GetBitContext *gb, AVCodecContext *avctx, H264ParamSets *ps, int bit_length)
Decode PPS.
Definition: h264_ps.c:696
MMCO_SET_MAX_LONG
@ MMCO_SET_MAX_LONG
Definition: h264_parse.h:64
AVCodecContext::framerate
AVRational framerate
Definition: avcodec.h:559
golomb.h
exp golomb vlc stuff
close
static av_cold void close(AVCodecParserContext *s)
Definition: apv_parser.c:197
get_bits
static unsigned int get_bits(GetBitContext *s, int n)
Read 1-25 bits.
Definition: get_bits.h:337
AV_FIELD_TT
@ AV_FIELD_TT
Top coded_first, top displayed first.
Definition: defs.h:214
ParseContext
Definition: parser.h:28
H264ParseContext::h264dsp
H264DSPContext h264dsp
Definition: h264_parser.c:58
fail
#define fail()
Definition: checkasm.h:218
ff_h264_sei_decode
int ff_h264_sei_decode(H264SEIContext *h, GetBitContext *gb, const H264ParamSets *ps, void *logctx)
Definition: h264_sei.c:230
H264_MAX_MMCO_COUNT
@ H264_MAX_MMCO_COUNT
Definition: h264.h:92
GetBitContext
Definition: get_bits.h:109
AV_PIX_FMT_YUV422P9
#define AV_PIX_FMT_YUV422P9
Definition: pixfmt.h:537
ff_h264_sei_process_picture_timing
int ff_h264_sei_process_picture_timing(H264SEIPictureTiming *h, const SPS *sps, void *logctx)
Parse the contents of a picture timing message given an active SPS.
Definition: h264_sei.c:64
ff_h264_golomb_to_pict_type
const uint8_t ff_h264_golomb_to_pict_type[5]
Definition: h264data.c:37
av_reduce
int av_reduce(int *dst_num, int *dst_den, int64_t num, int64_t den, int64_t max)
Reduce a fraction.
Definition: rational.c:35
AVRational::num
int num
Numerator.
Definition: rational.h:59
parse_nal_units
static int parse_nal_units(AVCodecParserContext *s, AVCodecContext *avctx, const uint8_t *const buf, int buf_size)
Parse NAL units of found picture and decode some basic information.
Definition: h264_parser.c:262
refstruct.h
h264_parse
static int h264_parse(AVCodecParserContext *s, AVCodecContext *avctx, const uint8_t **poutbuf, int *poutbuf_size, const uint8_t *buf, int buf_size)
Definition: h264_parser.c:589
AV_PIX_FMT_YUV444P10
#define AV_PIX_FMT_YUV444P10
Definition: pixfmt.h:542
ParseContext::buffer
uint8_t * buffer
Definition: parser.h:29
AV_PICTURE_STRUCTURE_FRAME
@ AV_PICTURE_STRUCTURE_FRAME
coded as frame
Definition: avcodec.h:2586
H264_NAL_DPA
@ H264_NAL_DPA
Definition: h264.h:36
AV_LOG_ERROR
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:210
av_cold
#define av_cold
Definition: attributes.h:106
init_get_bits8
static int init_get_bits8(GetBitContext *s, const uint8_t *buffer, int byte_size)
Initialize GetBitContext.
Definition: get_bits.h:544
AV_FIELD_UNKNOWN
@ AV_FIELD_UNKNOWN
Definition: defs.h:212
H2645NAL::size
int size
Definition: h2645_parse.h:36
AVCodecContext::extradata_size
int extradata_size
Definition: avcodec.h:523
s
#define s(width, name)
Definition: cbs_vp9.c:198
ff_h264_decode_extradata
int ff_h264_decode_extradata(const uint8_t *data, int size, H264ParamSets *ps, int *is_avc, int *nal_length_size, int err_recognition, void *logctx)
Definition: h264_parse.c:466
h264data.h
AV_PIX_FMT_YUV420P9
#define AV_PIX_FMT_YUV420P9
Definition: pixfmt.h:536
PICT_TOP_FIELD
#define PICT_TOP_FIELD
Definition: mpegutils.h:31
get_bits.h
H2645NAL::data
const uint8_t * data
Definition: h2645_parse.h:35
AV_PIX_FMT_YUV420P
@ AV_PIX_FMT_YUV420P
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:73
get_se_golomb
static int get_se_golomb(GetBitContext *gb)
read signed exp golomb code.
Definition: golomb.h:239
AV_CODEC_ID_H264
@ AV_CODEC_ID_H264
Definition: codec_id.h:79
H2645NAL::type
int type
NAL unit type.
Definition: h2645_parse.h:52
ff_h264_parser
const FFCodecParser ff_h264_parser
Definition: h264_parser.c:689
if
if(ret)
Definition: filter_design.txt:179
AV_PICTURE_STRUCTURE_BOTTOM_FIELD
@ AV_PICTURE_STRUCTURE_BOTTOM_FIELD
coded as bottom field
Definition: avcodec.h:2585
NULL
#define NULL
Definition: coverity.c:32
H264SEIContext
Definition: h264_sei.h:119
state
static struct @553 state
AVRational
Rational number (pair of numerator and denominator).
Definition: rational.h:58
SPS
Sequence parameter set.
Definition: h264_ps.h:44
AV_PICTURE_STRUCTURE_TOP_FIELD
@ AV_PICTURE_STRUCTURE_TOP_FIELD
coded as top field
Definition: avcodec.h:2584
H264_NAL_IDR_SLICE
@ H264_NAL_IDR_SLICE
Definition: h264.h:39
AV_PICTURE_TYPE_I
@ AV_PICTURE_TYPE_I
Intra.
Definition: avutil.h:278
get_bits1
static unsigned int get_bits1(GetBitContext *s)
Definition: get_bits.h:391
h264dsp.h
ParseContext::frame_start_found
int frame_start_found
Definition: parser.h:34
list
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining list
Definition: filter_design.txt:25
avpriv_find_start_code
const uint8_t * avpriv_find_start_code(const uint8_t *p, const uint8_t *end, uint32_t *state)
MAX_PPS_COUNT
#define MAX_PPS_COUNT
Definition: h264_ps.h:38
AV_PIX_FMT_YUV422P10
#define AV_PIX_FMT_YUV422P10
Definition: pixfmt.h:540
parse
static int parse(AVCodecParserContext *s, AVCodecContext *avctx, const uint8_t **poutbuf, int *poutbuf_size, const uint8_t *buf, int buf_size)
Definition: apv_parser.c:92
AVCodecContext::level
int level
Encoding level descriptor.
Definition: avcodec.h:1636
h264_find_frame_end
static int h264_find_frame_end(H264ParseContext *p, const uint8_t *buf, int buf_size, void *logctx)
Definition: h264_parser.c:82
h264_ps.h
index
int index
Definition: gxfenc.c:90
error.h
H264_SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM
@ H264_SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM
6: bottom field, top field, bottom field repeated, in that order
Definition: h264_sei.h:38
H2645RBSP::rbsp_buffer
uint8_t * rbsp_buffer
Definition: h2645_parse.h:75
startcode.h
H264DSPContext
Context for storing H.264 DSP functions.
Definition: h264dsp.h:42
av_sat_sub64
#define av_sat_sub64
Definition: common.h:142
MMCO_END
@ MMCO_END
Definition: h264_parse.h:60
i
#define i(width, name, range_min, range_max)
Definition: cbs_h264.c:63
AVCodecContext::pkt_timebase
AVRational pkt_timebase
Timebase in which pkt_dts/pts and AVPacket.dts/pts are expressed.
Definition: avcodec.h:550
AV_NOPTS_VALUE
#define AV_NOPTS_VALUE
Undefined timestamp value.
Definition: avutil.h:247
H264_SEI_PIC_STRUCT_BOTTOM_TOP
@ H264_SEI_PIC_STRUCT_BOTTOM_TOP
4: bottom field, top field, in that order
Definition: h264_sei.h:36
H2645NAL::gb
GetBitContext gb
Definition: h2645_parse.h:47
MMCO_SHORT2UNUSED
@ MMCO_SHORT2UNUSED
Definition: h264_parse.h:61
H2645NAL
Definition: h2645_parse.h:34
MMCO_RESET
@ MMCO_RESET
Definition: h264_parse.h:65
H264ParseContext::parse_history_count
int parse_history_count
Definition: h264_parser.c:66
H264ParseContext::last_frame_num
int last_frame_num
Definition: h264_parser.c:69
attributes.h
mb
#define mb
Definition: vf_colormatrix.c:99
ff_combine_frame
int ff_combine_frame(ParseContext *pc, int next, const uint8_t **buf, int *buf_size)
Combine the (truncated) bitstream to a complete frame.
Definition: parser.c:211
FFCodecParser
Definition: parser_internal.h:29
H264_SEI_PIC_STRUCT_FRAME
@ H264_SEI_PIC_STRUCT_FRAME
0: frame
Definition: h264_sei.h:32
H264_SEI_PIC_STRUCT_FRAME_TRIPLING
@ H264_SEI_PIC_STRUCT_FRAME_TRIPLING
8: frame tripling
Definition: h264_sei.h:40
h264_sei.h
ff_h264_decode_seq_parameter_set
int ff_h264_decode_seq_parameter_set(GetBitContext *gb, AVCodecContext *avctx, H264ParamSets *ps, int ignore_truncation)
Decode SPS.
Definition: h264_ps.c:284
log.h
PARSER_FLAG_COMPLETE_FRAMES
#define PARSER_FLAG_COMPLETE_FRAMES
Definition: avcodec.h:2623
MMCO_SHORT2LONG
@ MMCO_SHORT2LONG
Definition: h264_parse.h:63
H264_SEI_PIC_STRUCT_FRAME_DOUBLING
@ H264_SEI_PIC_STRUCT_FRAME_DOUBLING
7: frame doubling
Definition: h264_sei.h:39
AVCodecContext::extradata
uint8_t * extradata
Out-of-band global headers that may be used by some codecs.
Definition: avcodec.h:522
ff_h2645_extract_rbsp
int ff_h2645_extract_rbsp(const uint8_t *src, int length, H2645RBSP *rbsp, H2645NAL *nal, int small_padding)
Extract the raw (unescaped) bitstream.
Definition: h2645_parse.c:37
H264POCContext
Definition: h264_parse.h:82
AV_FIELD_BB
@ AV_FIELD_BB
Bottom coded first, bottom displayed first.
Definition: defs.h:215
av_assert1
#define av_assert1(cond)
assert() equivalent, that does not lie in speed critical code.
Definition: avassert.h:58
av_fast_padded_malloc
void av_fast_padded_malloc(void *ptr, unsigned int *size, size_t min_size)
Same behaviour av_fast_malloc but the buffer has additional AV_INPUT_BUFFER_PADDING_SIZE at the end w...
Definition: utils.c:53
FFMIN
#define FFMIN(a, b)
Definition: macros.h:49
av_inv_q
static av_always_inline AVRational av_inv_q(AVRational q)
Invert a rational.
Definition: rational.h:159
parser.h
ff_h264_parse_ref_count
int ff_h264_parse_ref_count(int *plist_count, int ref_count[2], GetBitContext *gb, const PPS *pps, int slice_type_nos, int picture_structure, void *logctx)
Definition: h264_parse.c:222
av_rescale
int64_t av_rescale(int64_t a, int64_t b, int64_t c)
Rescale a 64-bit integer with rounding to nearest.
Definition: mathematics.c:129
H264ParseContext::last_picture_structure
int last_picture_structure
Definition: h264_parser.c:69
PARSER_CODEC_LIST
#define PARSER_CODEC_LIST(...)
Definition: parser_internal.h:76
AV_PIX_FMT_YUV444P9
#define AV_PIX_FMT_YUV444P9
Definition: pixfmt.h:538
H2645RBSP::rbsp_buffer_alloc_size
int rbsp_buffer_alloc_size
Definition: h2645_parse.h:77
avcodec.h
AVCodecParserContext
Definition: avcodec.h:2589
ff_h264dsp_init
av_cold void ff_h264dsp_init(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)
Definition: h264dsp.c:66
ret
ret
Definition: filter_design.txt:187
ff_h264_init_poc
int ff_h264_init_poc(int pic_field_poc[2], int *pic_poc, const SPS *sps, H264POCContext *pc, int picture_structure, int nal_ref_idc)
Definition: h264_parse.c:280
pixfmt.h
ff_h264_get_profile
int ff_h264_get_profile(const SPS *sps)
Compute profile from profile_idc and constraint_set?_flags.
Definition: h264_parse.c:533
H264_NAL_SLICE
@ H264_NAL_SLICE
Definition: h264.h:35
sps
static int FUNC() sps(CodedBitstreamContext *ctx, RWContext *rw, H264RawSPS *current)
Definition: cbs_h264_syntax_template.c:260
av_sat_add64
#define av_sat_add64
Definition: common.h:139
H264_NAL_SEI
@ H264_NAL_SEI
Definition: h264.h:40
U
#define U(x)
Definition: vpx_arith.h:37
H264_NAL_SPS
@ H264_NAL_SPS
Definition: h264.h:41
AVCodecContext
main external API structure.
Definition: avcodec.h:439
get_ue_golomb_31
static int get_ue_golomb_31(GetBitContext *gb)
read unsigned exp golomb code, constraint to a max of 31.
Definition: golomb.h:120
find_start_code
static int find_start_code(const uint8_t *buf, int buf_size, int buf_index, int next_avc)
Definition: h264_parser.c:72
AV_PICTURE_TYPE_B
@ AV_PICTURE_TYPE_B
Bi-dir predicted.
Definition: avutil.h:280
av_refstruct_replace
void av_refstruct_replace(void *dstp, const void *src)
Ensure *dstp refers to the same object as src.
Definition: refstruct.c:160
AVRational::den
int den
Denominator.
Definition: rational.h:60
AV_PIX_FMT_NONE
@ AV_PIX_FMT_NONE
Definition: pixfmt.h:72
H264ParseContext::nal_length_size
int nal_length_size
Definition: h264_parser.c:62
AVCodecContext::profile
int profile
profile
Definition: avcodec.h:1626
H264ParamSets
Definition: h264_ps.h:144
Windows::Graphics::DirectX::Direct3D11::p
IDirect3DDxgiInterfaceAccess _COM_Outptr_ void ** p
Definition: vsrc_gfxcapture_winrt.hpp:53
av_mul_q
AVRational av_mul_q(AVRational b, AVRational c)
Multiply two rationals.
Definition: rational.c:80
H264PredWeightTable
Definition: h264_parse.h:69
AV_PIX_FMT_YUV444P
@ AV_PIX_FMT_YUV444P
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
Definition: pixfmt.h:78
AV_PICTURE_TYPE_P
@ AV_PICTURE_TYPE_P
Predicted.
Definition: avutil.h:279
avutil.h
AV_PIX_FMT_YUV422P
@ AV_PIX_FMT_YUV422P
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Definition: pixfmt.h:77
mem.h
init
static av_cold int init(AVCodecParserContext *s)
Definition: h264_parser.c:679
get_ue_golomb_long
static unsigned get_ue_golomb_long(GetBitContext *gb)
Read an unsigned Exp-Golomb code in the range 0 to UINT32_MAX-1.
Definition: golomb.h:104
H264_SEI_PIC_STRUCT_BOTTOM_FIELD
@ H264_SEI_PIC_STRUCT_BOTTOM_FIELD
2: bottom field
Definition: h264_sei.h:34
H264ParseContext::ps
H264ParamSets ps
Definition: h264_parser.c:57
H264ParseContext::poc
H264POCContext poc
Definition: h264_parser.c:59
END_NOT_FOUND
#define END_NOT_FOUND
Definition: parser.h:40
av_freep
#define av_freep(p)
Definition: tableprint_vlc.h:35
h264.h
get_nalsize
static int get_nalsize(int nal_length_size, const uint8_t *buf, int buf_size, int *buf_index, void *logctx)
Definition: h2645_parse.h:126
H264ParseContext::reference_dts
int64_t reference_dts
Definition: h264_parser.c:68
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:27
AVERROR_INVALIDDATA
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:61
H264_NAL_AUD
@ H264_NAL_AUD
Definition: h264.h:43
PARSER_FLAG_ONCE
#define PARSER_FLAG_ONCE
Definition: avcodec.h:2624
H2645RBSP
Definition: h2645_parse.h:74
H264_SEI_PIC_STRUCT_TOP_FIELD
@ H264_SEI_PIC_STRUCT_TOP_FIELD
1: top field
Definition: h264_sei.h:33
H264_SEI_PIC_STRUCT_TOP_BOTTOM_TOP
@ H264_SEI_PIC_STRUCT_TOP_BOTTOM_TOP
5: top field, bottom field, top field repeated, in that order
Definition: h264_sei.h:37
H264ParseContext::pc
ParseContext pc
Definition: h264_parser.c:56