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 / MPEG4 part10 parser.
25 * @author Michael Niedermayer <michaelni@gmx.at>
26 */
27
28 #define UNCHECKED_BITSTREAM_READER 1
29
30 #include "libavutil/attributes.h"
31 #include "parser.h"
32 #include "h264data.h"
33 #include "golomb.h"
34 #include "internal.h"
35 #include "mpegutils.h"
36
37
h264_find_frame_end(H264Context * h,const uint8_t * buf,int buf_size)38 static int h264_find_frame_end(H264Context *h, const uint8_t *buf,
39 int buf_size)
40 {
41 int i, j;
42 uint32_t state;
43 ParseContext *pc = &h->parse_context;
44 int next_avc= h->is_avc ? 0 : buf_size;
45
46 // mb_addr= pc->mb_addr - 1;
47 state = pc->state;
48 if (state > 13)
49 state = 7;
50
51 if (h->is_avc && !h->nal_length_size)
52 av_log(h->avctx, AV_LOG_ERROR, "AVC-parser: nal length size invalid\n");
53
54 for (i = 0; i < buf_size; i++) {
55 if (i >= next_avc) {
56 int nalsize = 0;
57 i = next_avc;
58 for (j = 0; j < h->nal_length_size; j++)
59 nalsize = (nalsize << 8) | buf[i++];
60 if (nalsize <= 0 || nalsize > buf_size - i) {
61 av_log(h->avctx, AV_LOG_ERROR, "AVC-parser: nal size %d remaining %d\n", nalsize, buf_size - i);
62 return buf_size;
63 }
64 next_avc = i + nalsize;
65 state = 5;
66 }
67
68 if (state == 7) {
69 i += h->h264dsp.startcode_find_candidate(buf + i, next_avc - i);
70 if (i < next_avc)
71 state = 2;
72 } else if (state <= 2) {
73 if (buf[i] == 1)
74 state ^= 5; // 2->7, 1->4, 0->5
75 else if (buf[i])
76 state = 7;
77 else
78 state >>= 1; // 2->1, 1->0, 0->0
79 } else if (state <= 5) {
80 int nalu_type = buf[i] & 0x1F;
81 if (nalu_type == NAL_SEI || nalu_type == NAL_SPS ||
82 nalu_type == NAL_PPS || nalu_type == NAL_AUD) {
83 if (pc->frame_start_found) {
84 i++;
85 goto found;
86 }
87 } else if (nalu_type == NAL_SLICE || nalu_type == NAL_DPA ||
88 nalu_type == NAL_IDR_SLICE) {
89 state += 8;
90 continue;
91 }
92 state = 7;
93 } else {
94 h->parse_history[h->parse_history_count++]= buf[i];
95 if (h->parse_history_count>5) {
96 unsigned int mb, last_mb= h->parse_last_mb;
97 GetBitContext gb;
98
99 init_get_bits(&gb, h->parse_history, 8*h->parse_history_count);
100 h->parse_history_count=0;
101 mb= get_ue_golomb_long(&gb);
102 h->parse_last_mb= mb;
103 if (pc->frame_start_found) {
104 if (mb <= last_mb)
105 goto found;
106 } else
107 pc->frame_start_found = 1;
108 state = 7;
109 }
110 }
111 }
112 pc->state = state;
113 if (h->is_avc)
114 return next_avc;
115 return END_NOT_FOUND;
116
117 found:
118 pc->state = 7;
119 pc->frame_start_found = 0;
120 if (h->is_avc)
121 return next_avc;
122 return i - (state & 5) - 5 * (state > 7);
123 }
124
scan_mmco_reset(AVCodecParserContext * s)125 static int scan_mmco_reset(AVCodecParserContext *s)
126 {
127 H264Context *h = s->priv_data;
128
129 h->slice_type_nos = s->pict_type & 3;
130
131 if (h->pps.redundant_pic_cnt_present)
132 get_ue_golomb(&h->gb); // redundant_pic_count
133
134 if (ff_set_ref_count(h) < 0)
135 return AVERROR_INVALIDDATA;
136
137 if (h->slice_type_nos != AV_PICTURE_TYPE_I) {
138 int list;
139 for (list = 0; list < h->list_count; list++) {
140 if (get_bits1(&h->gb)) {
141 int index;
142 for (index = 0; ; index++) {
143 unsigned int reordering_of_pic_nums_idc = get_ue_golomb_31(&h->gb);
144
145 if (reordering_of_pic_nums_idc < 3)
146 get_ue_golomb(&h->gb);
147 else if (reordering_of_pic_nums_idc > 3) {
148 av_log(h->avctx, AV_LOG_ERROR,
149 "illegal reordering_of_pic_nums_idc %d\n",
150 reordering_of_pic_nums_idc);
151 return AVERROR_INVALIDDATA;
152 } else
153 break;
154
155 if (index >= h->ref_count[list]) {
156 av_log(h->avctx, AV_LOG_ERROR,
157 "reference count %d overflow\n", index);
158 return AVERROR_INVALIDDATA;
159 }
160 }
161 }
162 }
163 }
164
165 if ((h->pps.weighted_pred && h->slice_type_nos == AV_PICTURE_TYPE_P) ||
166 (h->pps.weighted_bipred_idc == 1 && h->slice_type_nos == AV_PICTURE_TYPE_B))
167 ff_pred_weight_table(h);
168
169 if (get_bits1(&h->gb)) { // adaptive_ref_pic_marking_mode_flag
170 int i;
171 for (i = 0; i < MAX_MMCO_COUNT; i++) {
172 MMCOOpcode opcode = get_ue_golomb_31(&h->gb);
173 if (opcode > (unsigned) MMCO_LONG) {
174 av_log(h->avctx, AV_LOG_ERROR,
175 "illegal memory management control operation %d\n",
176 opcode);
177 return AVERROR_INVALIDDATA;
178 }
179 if (opcode == MMCO_END)
180 return 0;
181 else if (opcode == MMCO_RESET)
182 return 1;
183
184 if (opcode == MMCO_SHORT2UNUSED || opcode == MMCO_SHORT2LONG)
185 get_ue_golomb(&h->gb);
186 if (opcode == MMCO_SHORT2LONG || opcode == MMCO_LONG2UNUSED ||
187 opcode == MMCO_LONG || opcode == MMCO_SET_MAX_LONG)
188 get_ue_golomb_31(&h->gb);
189 }
190 }
191
192 return 0;
193 }
194
195 /**
196 * Parse NAL units of found picture and decode some basic information.
197 *
198 * @param s parser context.
199 * @param avctx codec context.
200 * @param buf buffer with field/frame data.
201 * @param buf_size size of the buffer.
202 */
parse_nal_units(AVCodecParserContext * s,AVCodecContext * avctx,const uint8_t * buf,int buf_size)203 static inline int parse_nal_units(AVCodecParserContext *s,
204 AVCodecContext *avctx,
205 const uint8_t *buf, int buf_size)
206 {
207 H264Context *h = s->priv_data;
208 const uint8_t *buf_end = buf + buf_size;
209 unsigned int pps_id;
210 unsigned int slice_type;
211 int state = -1, got_reset = 0;
212 const uint8_t *ptr;
213 int q264 = buf_size >=4 && !memcmp("Q264", buf, 4);
214 int field_poc[2];
215
216 /* set some sane default values */
217 s->pict_type = AV_PICTURE_TYPE_I;
218 s->key_frame = 0;
219 s->picture_structure = AV_PICTURE_STRUCTURE_UNKNOWN;
220
221 h->avctx = avctx;
222 ff_h264_reset_sei(h);
223 h->sei_fpa.frame_packing_arrangement_cancel_flag = -1;
224
225 if (!buf_size)
226 return 0;
227
228 for (;;) {
229 int src_length, dst_length, consumed, nalsize = 0;
230 if (h->is_avc) {
231 int i;
232 if (h->nal_length_size >= buf_end - buf) break;
233 nalsize = 0;
234 for (i = 0; i < h->nal_length_size; i++)
235 nalsize = (nalsize << 8) | *buf++;
236 if (nalsize <= 0 || nalsize > buf_end - buf) {
237 av_log(h->avctx, AV_LOG_ERROR, "AVC: nal size %d\n", nalsize);
238 break;
239 }
240 src_length = nalsize;
241 } else {
242 buf = avpriv_find_start_code(buf, buf_end, &state);
243 if (buf >= buf_end)
244 break;
245 --buf;
246 src_length = buf_end - buf;
247 }
248 switch (state & 0x1f) {
249 case NAL_SLICE:
250 case NAL_IDR_SLICE:
251 // Do not walk the whole buffer just to decode slice header
252 if ((state & 0x1f) == NAL_IDR_SLICE || ((state >> 5) & 0x3) == 0) {
253 /* IDR or disposable slice
254 * No need to decode many bytes because MMCOs shall not be present. */
255 if (src_length > 60)
256 src_length = 60;
257 } else {
258 /* To decode up to MMCOs */
259 if (src_length > 1000)
260 src_length = 1000;
261 }
262 break;
263 }
264 ptr = ff_h264_decode_nal(h, buf, &dst_length, &consumed, src_length);
265 if (!ptr || dst_length < 0)
266 break;
267
268 init_get_bits(&h->gb, ptr, 8 * dst_length);
269 switch (h->nal_unit_type) {
270 case NAL_SPS:
271 ff_h264_decode_seq_parameter_set(h);
272 break;
273 case NAL_PPS:
274 ff_h264_decode_picture_parameter_set(h, h->gb.size_in_bits);
275 break;
276 case NAL_SEI:
277 ff_h264_decode_sei(h);
278 break;
279 case NAL_IDR_SLICE:
280 s->key_frame = 1;
281
282 h->prev_frame_num = 0;
283 h->prev_frame_num_offset = 0;
284 h->prev_poc_msb =
285 h->prev_poc_lsb = 0;
286 /* fall through */
287 case NAL_SLICE:
288 get_ue_golomb_long(&h->gb); // skip first_mb_in_slice
289 slice_type = get_ue_golomb_31(&h->gb);
290 s->pict_type = golomb_to_pict_type[slice_type % 5];
291 if (h->sei_recovery_frame_cnt >= 0) {
292 /* key frame, since recovery_frame_cnt is set */
293 s->key_frame = 1;
294 }
295 pps_id = get_ue_golomb(&h->gb);
296 if (pps_id >= MAX_PPS_COUNT) {
297 av_log(h->avctx, AV_LOG_ERROR,
298 "pps_id %u out of range\n", pps_id);
299 return -1;
300 }
301 if (!h->pps_buffers[pps_id]) {
302 av_log(h->avctx, AV_LOG_ERROR,
303 "non-existing PPS %u referenced\n", pps_id);
304 return -1;
305 }
306 h->pps = *h->pps_buffers[pps_id];
307 if (!h->sps_buffers[h->pps.sps_id]) {
308 av_log(h->avctx, AV_LOG_ERROR,
309 "non-existing SPS %u referenced\n", h->pps.sps_id);
310 return -1;
311 }
312 h->sps = *h->sps_buffers[h->pps.sps_id];
313 h->frame_num = get_bits(&h->gb, h->sps.log2_max_frame_num);
314
315 if(h->sps.ref_frame_count <= 1 && h->pps.ref_count[0] <= 1 && s->pict_type == AV_PICTURE_TYPE_I)
316 s->key_frame = 1;
317
318 avctx->profile = ff_h264_get_profile(&h->sps);
319 avctx->level = h->sps.level_idc;
320
321 if (h->sps.frame_mbs_only_flag) {
322 h->picture_structure = PICT_FRAME;
323 } else {
324 if (get_bits1(&h->gb)) { // field_pic_flag
325 h->picture_structure = PICT_TOP_FIELD + get_bits1(&h->gb); // bottom_field_flag
326 } else {
327 h->picture_structure = PICT_FRAME;
328 }
329 }
330
331 if (h->nal_unit_type == NAL_IDR_SLICE)
332 get_ue_golomb(&h->gb); /* idr_pic_id */
333 if (h->sps.poc_type == 0) {
334 h->poc_lsb = get_bits(&h->gb, h->sps.log2_max_poc_lsb);
335
336 if (h->pps.pic_order_present == 1 &&
337 h->picture_structure == PICT_FRAME)
338 h->delta_poc_bottom = get_se_golomb(&h->gb);
339 }
340
341 if (h->sps.poc_type == 1 &&
342 !h->sps.delta_pic_order_always_zero_flag) {
343 h->delta_poc[0] = get_se_golomb(&h->gb);
344
345 if (h->pps.pic_order_present == 1 &&
346 h->picture_structure == PICT_FRAME)
347 h->delta_poc[1] = get_se_golomb(&h->gb);
348 }
349
350 /* Decode POC of this picture.
351 * The prev_ values needed for decoding POC of the next picture are not set here. */
352 field_poc[0] = field_poc[1] = INT_MAX;
353 ff_init_poc(h, field_poc, &s->output_picture_number);
354
355 /* Continue parsing to check if MMCO_RESET is present.
356 * FIXME: MMCO_RESET could appear in non-first slice.
357 * Maybe, we should parse all undisposable non-IDR slice of this
358 * picture until encountering MMCO_RESET in a slice of it. */
359 if (h->nal_ref_idc && h->nal_unit_type != NAL_IDR_SLICE) {
360 got_reset = scan_mmco_reset(s);
361 if (got_reset < 0)
362 return got_reset;
363 }
364
365 /* Set up the prev_ values for decoding POC of the next picture. */
366 h->prev_frame_num = got_reset ? 0 : h->frame_num;
367 h->prev_frame_num_offset = got_reset ? 0 : h->frame_num_offset;
368 if (h->nal_ref_idc != 0) {
369 if (!got_reset) {
370 h->prev_poc_msb = h->poc_msb;
371 h->prev_poc_lsb = h->poc_lsb;
372 } else {
373 h->prev_poc_msb = 0;
374 h->prev_poc_lsb =
375 h->picture_structure == PICT_BOTTOM_FIELD ? 0 : field_poc[0];
376 }
377 }
378
379 if (h->sps.pic_struct_present_flag) {
380 switch (h->sei_pic_struct) {
381 case SEI_PIC_STRUCT_TOP_FIELD:
382 case SEI_PIC_STRUCT_BOTTOM_FIELD:
383 s->repeat_pict = 0;
384 break;
385 case SEI_PIC_STRUCT_FRAME:
386 case SEI_PIC_STRUCT_TOP_BOTTOM:
387 case SEI_PIC_STRUCT_BOTTOM_TOP:
388 s->repeat_pict = 1;
389 break;
390 case SEI_PIC_STRUCT_TOP_BOTTOM_TOP:
391 case SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM:
392 s->repeat_pict = 2;
393 break;
394 case SEI_PIC_STRUCT_FRAME_DOUBLING:
395 s->repeat_pict = 3;
396 break;
397 case SEI_PIC_STRUCT_FRAME_TRIPLING:
398 s->repeat_pict = 5;
399 break;
400 default:
401 s->repeat_pict = h->picture_structure == PICT_FRAME ? 1 : 0;
402 break;
403 }
404 } else {
405 s->repeat_pict = h->picture_structure == PICT_FRAME ? 1 : 0;
406 }
407
408 if (h->picture_structure == PICT_FRAME) {
409 s->picture_structure = AV_PICTURE_STRUCTURE_FRAME;
410 if (h->sps.pic_struct_present_flag) {
411 switch (h->sei_pic_struct) {
412 case SEI_PIC_STRUCT_TOP_BOTTOM:
413 case SEI_PIC_STRUCT_TOP_BOTTOM_TOP:
414 s->field_order = AV_FIELD_TT;
415 break;
416 case SEI_PIC_STRUCT_BOTTOM_TOP:
417 case SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM:
418 s->field_order = AV_FIELD_BB;
419 break;
420 default:
421 s->field_order = AV_FIELD_PROGRESSIVE;
422 break;
423 }
424 } else {
425 if (field_poc[0] < field_poc[1])
426 s->field_order = AV_FIELD_TT;
427 else if (field_poc[0] > field_poc[1])
428 s->field_order = AV_FIELD_BB;
429 else
430 s->field_order = AV_FIELD_PROGRESSIVE;
431 }
432 } else {
433 if (h->picture_structure == PICT_TOP_FIELD)
434 s->picture_structure = AV_PICTURE_STRUCTURE_TOP_FIELD;
435 else
436 s->picture_structure = AV_PICTURE_STRUCTURE_BOTTOM_FIELD;
437 s->field_order = AV_FIELD_UNKNOWN;
438 }
439
440 return 0; /* no need to evaluate the rest */
441 }
442 buf += h->is_avc ? nalsize : consumed;
443 }
444 if (q264)
445 return 0;
446 /* didn't find a picture! */
447 av_log(h->avctx, AV_LOG_ERROR, "missing picture in access unit with size %d\n", buf_size);
448 return -1;
449 }
450
h264_parse(AVCodecParserContext * s,AVCodecContext * avctx,const uint8_t ** poutbuf,int * poutbuf_size,const uint8_t * buf,int buf_size)451 static int h264_parse(AVCodecParserContext *s,
452 AVCodecContext *avctx,
453 const uint8_t **poutbuf, int *poutbuf_size,
454 const uint8_t *buf, int buf_size)
455 {
456 H264Context *h = s->priv_data;
457 ParseContext *pc = &h->parse_context;
458 int next;
459
460 if (!h->got_first) {
461 h->got_first = 1;
462 if (avctx->extradata_size) {
463 h->avctx = avctx;
464 // must be done like in decoder, otherwise opening the parser,
465 // letting it create extradata and then closing and opening again
466 // will cause has_b_frames to be always set.
467 // Note that estimate_timings_from_pts does exactly this.
468 if (!avctx->has_b_frames)
469 h->low_delay = 1;
470 ff_h264_decode_extradata(h, avctx->extradata, avctx->extradata_size);
471 }
472 }
473
474 if (s->flags & PARSER_FLAG_COMPLETE_FRAMES) {
475 next = buf_size;
476 } else {
477 next = h264_find_frame_end(h, buf, buf_size);
478
479 if (ff_combine_frame(pc, next, &buf, &buf_size) < 0) {
480 *poutbuf = NULL;
481 *poutbuf_size = 0;
482 return buf_size;
483 }
484
485 if (next < 0 && next != END_NOT_FOUND) {
486 av_assert1(pc->last_index + next >= 0);
487 h264_find_frame_end(h, &pc->buffer[pc->last_index + next], -next); // update state
488 }
489 }
490
491 parse_nal_units(s, avctx, buf, buf_size);
492
493 if (h->sei_cpb_removal_delay >= 0) {
494 s->dts_sync_point = h->sei_buffering_period_present;
495 s->dts_ref_dts_delta = h->sei_cpb_removal_delay;
496 s->pts_dts_delta = h->sei_dpb_output_delay;
497 } else {
498 s->dts_sync_point = INT_MIN;
499 s->dts_ref_dts_delta = INT_MIN;
500 s->pts_dts_delta = INT_MIN;
501 }
502
503 if (s->flags & PARSER_FLAG_ONCE) {
504 s->flags &= PARSER_FLAG_COMPLETE_FRAMES;
505 }
506
507 *poutbuf = buf;
508 *poutbuf_size = buf_size;
509 return next;
510 }
511
h264_split(AVCodecContext * avctx,const uint8_t * buf,int buf_size)512 static int h264_split(AVCodecContext *avctx,
513 const uint8_t *buf, int buf_size)
514 {
515 int i;
516 uint32_t state = -1;
517 int has_sps = 0;
518
519 for (i = 0; i <= buf_size; i++) {
520 if ((state & 0xFFFFFF1F) == 0x107)
521 has_sps = 1;
522 /* if ((state&0xFFFFFF1F) == 0x101 ||
523 * (state&0xFFFFFF1F) == 0x102 ||
524 * (state&0xFFFFFF1F) == 0x105) {
525 * }
526 */
527 if ((state & 0xFFFFFF00) == 0x100 && (state & 0xFFFFFF1F) != 0x107 &&
528 (state & 0xFFFFFF1F) != 0x108 && (state & 0xFFFFFF1F) != 0x109) {
529 if (has_sps) {
530 while (i > 4 && buf[i - 5] == 0)
531 i--;
532 return i - 4;
533 }
534 }
535 if (i < buf_size)
536 state = (state << 8) | buf[i];
537 }
538 return 0;
539 }
540
close(AVCodecParserContext * s)541 static void close(AVCodecParserContext *s)
542 {
543 H264Context *h = s->priv_data;
544 ParseContext *pc = &h->parse_context;
545
546 av_free(pc->buffer);
547 ff_h264_free_context(h);
548 }
549
init(AVCodecParserContext * s)550 static av_cold int init(AVCodecParserContext *s)
551 {
552 H264Context *h = s->priv_data;
553 h->thread_context[0] = h;
554 h->slice_context_count = 1;
555 ff_h264dsp_init(&h->h264dsp, 8, 1);
556 return 0;
557 }
558
559 AVCodecParser ff_h264_parser = {
560 .codec_ids = { AV_CODEC_ID_H264 },
561 .priv_data_size = sizeof(H264Context),
562 .parser_init = init,
563 .parser_parse = h264_parse,
564 .parser_close = close,
565 .split = h264_split,
566 };
567