1 /*
2 * HEVC Annex B format parser
3 *
4 * Copyright (C) 2012 - 2013 Guillaume Martres
5 *
6 * This file is part of FFmpeg.
7 *
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23 #include "libavutil/common.h"
24
25 #include "golomb.h"
26 #include "hevc.h"
27 #include "hevc_parse.h"
28 #include "hevc_ps.h"
29 #include "hevc_sei.h"
30 #include "h2645_parse.h"
31 #include "internal.h"
32 #include "parser.h"
33
34 #define START_CODE 0x000001 ///< start_code_prefix_one_3bytes
35
36 #define IS_IRAP_NAL(nal) (nal->type >= 16 && nal->type <= 23)
37 #define IS_IDR_NAL(nal) (nal->type == HEVC_NAL_IDR_W_RADL || nal->type == HEVC_NAL_IDR_N_LP)
38
39 typedef struct HEVCParserContext {
40 ParseContext pc;
41
42 H2645Packet pkt;
43 HEVCParamSets ps;
44 HEVCSEI sei;
45
46 int is_avc;
47 int nal_length_size;
48 int parsed_extradata;
49
50 int poc;
51 int pocTid0;
52 } HEVCParserContext;
53
hevc_parse_slice_header(AVCodecParserContext * s,H2645NAL * nal,AVCodecContext * avctx)54 static int hevc_parse_slice_header(AVCodecParserContext *s, H2645NAL *nal,
55 AVCodecContext *avctx)
56 {
57 HEVCParserContext *ctx = s->priv_data;
58 HEVCParamSets *ps = &ctx->ps;
59 HEVCSEI *sei = &ctx->sei;
60 GetBitContext *gb = &nal->gb;
61 const HEVCWindow *ow;
62 int i, num = 0, den = 0;
63
64 unsigned int pps_id, first_slice_in_pic_flag, dependent_slice_segment_flag;
65 enum HEVCSliceType slice_type;
66
67 first_slice_in_pic_flag = get_bits1(gb);
68 s->picture_structure = sei->picture_timing.picture_struct;
69 s->field_order = sei->picture_timing.picture_struct;
70
71 if (IS_IRAP_NAL(nal)) {
72 s->key_frame = 1;
73 skip_bits1(gb); // no_output_of_prior_pics_flag
74 }
75
76 pps_id = get_ue_golomb(gb);
77 if (pps_id >= HEVC_MAX_PPS_COUNT || !ps->pps_list[pps_id]) {
78 av_log(avctx, AV_LOG_ERROR, "PPS id out of range: %d\n", pps_id);
79 return AVERROR_INVALIDDATA;
80 }
81 ps->pps = (HEVCPPS*)ps->pps_list[pps_id]->data;
82
83 if (ps->pps->sps_id >= HEVC_MAX_SPS_COUNT || !ps->sps_list[ps->pps->sps_id]) {
84 av_log(avctx, AV_LOG_ERROR, "SPS id out of range: %d\n", ps->pps->sps_id);
85 return AVERROR_INVALIDDATA;
86 }
87 if (ps->sps != (HEVCSPS*)ps->sps_list[ps->pps->sps_id]->data) {
88 ps->sps = (HEVCSPS*)ps->sps_list[ps->pps->sps_id]->data;
89 ps->vps = (HEVCVPS*)ps->vps_list[ps->sps->vps_id]->data;
90 }
91 ow = &ps->sps->output_window;
92
93 s->coded_width = ps->sps->width;
94 s->coded_height = ps->sps->height;
95 s->width = ps->sps->width - ow->left_offset - ow->right_offset;
96 s->height = ps->sps->height - ow->top_offset - ow->bottom_offset;
97 s->format = ps->sps->pix_fmt;
98 avctx->profile = ps->sps->ptl.general_ptl.profile_idc;
99 avctx->level = ps->sps->ptl.general_ptl.level_idc;
100
101 if (ps->vps->vps_timing_info_present_flag) {
102 num = ps->vps->vps_num_units_in_tick;
103 den = ps->vps->vps_time_scale;
104 } else if (ps->sps->vui.vui_timing_info_present_flag) {
105 num = ps->sps->vui.vui_num_units_in_tick;
106 den = ps->sps->vui.vui_time_scale;
107 }
108
109 if (num != 0 && den != 0)
110 av_reduce(&avctx->framerate.den, &avctx->framerate.num,
111 num, den, 1 << 30);
112
113 if (!first_slice_in_pic_flag) {
114 unsigned int slice_segment_addr;
115 int slice_address_length;
116
117 if (ps->pps->dependent_slice_segments_enabled_flag)
118 dependent_slice_segment_flag = get_bits1(gb);
119 else
120 dependent_slice_segment_flag = 0;
121
122 slice_address_length = av_ceil_log2_c(ps->sps->ctb_width *
123 ps->sps->ctb_height);
124 slice_segment_addr = get_bitsz(gb, slice_address_length);
125 if (slice_segment_addr >= ps->sps->ctb_width * ps->sps->ctb_height) {
126 av_log(avctx, AV_LOG_ERROR, "Invalid slice segment address: %u.\n",
127 slice_segment_addr);
128 return AVERROR_INVALIDDATA;
129 }
130 } else
131 dependent_slice_segment_flag = 0;
132
133 if (dependent_slice_segment_flag)
134 return 0; /* break; */
135
136 for (i = 0; i < ps->pps->num_extra_slice_header_bits; i++)
137 skip_bits(gb, 1); // slice_reserved_undetermined_flag[]
138
139 slice_type = get_ue_golomb_31(gb);
140 if (!(slice_type == HEVC_SLICE_I || slice_type == HEVC_SLICE_P ||
141 slice_type == HEVC_SLICE_B)) {
142 av_log(avctx, AV_LOG_ERROR, "Unknown slice type: %d.\n",
143 slice_type);
144 return AVERROR_INVALIDDATA;
145 }
146 s->pict_type = slice_type == HEVC_SLICE_B ? AV_PICTURE_TYPE_B :
147 slice_type == HEVC_SLICE_P ? AV_PICTURE_TYPE_P :
148 AV_PICTURE_TYPE_I;
149
150 if (ps->pps->output_flag_present_flag)
151 skip_bits1(gb); // pic_output_flag
152
153 if (ps->sps->separate_colour_plane_flag)
154 skip_bits(gb, 2); // colour_plane_id
155
156 if (!IS_IDR_NAL(nal)) {
157 int pic_order_cnt_lsb = get_bits(gb, ps->sps->log2_max_poc_lsb);
158 s->output_picture_number = ctx->poc =
159 ff_hevc_compute_poc(ps->sps, ctx->pocTid0, pic_order_cnt_lsb, nal->type);
160 } else
161 s->output_picture_number = ctx->poc = 0;
162
163 if (nal->temporal_id == 0 &&
164 nal->type != HEVC_NAL_TRAIL_N &&
165 nal->type != HEVC_NAL_TSA_N &&
166 nal->type != HEVC_NAL_STSA_N &&
167 nal->type != HEVC_NAL_RADL_N &&
168 nal->type != HEVC_NAL_RASL_N &&
169 nal->type != HEVC_NAL_RADL_R &&
170 nal->type != HEVC_NAL_RASL_R)
171 ctx->pocTid0 = ctx->poc;
172
173 return 1; /* no need to evaluate the rest */
174 }
175
176 /**
177 * Parse NAL units of found picture and decode some basic information.
178 *
179 * @param s parser context.
180 * @param avctx codec context.
181 * @param buf buffer with field/frame data.
182 * @param buf_size size of the buffer.
183 */
parse_nal_units(AVCodecParserContext * s,const uint8_t * buf,int buf_size,AVCodecContext * avctx)184 static int parse_nal_units(AVCodecParserContext *s, const uint8_t *buf,
185 int buf_size, AVCodecContext *avctx)
186 {
187 HEVCParserContext *ctx = s->priv_data;
188 HEVCParamSets *ps = &ctx->ps;
189 HEVCSEI *sei = &ctx->sei;
190 int ret, i;
191
192 /* set some sane default values */
193 s->pict_type = AV_PICTURE_TYPE_I;
194 s->key_frame = 0;
195 s->picture_structure = AV_PICTURE_STRUCTURE_UNKNOWN;
196
197 ff_hevc_reset_sei(sei);
198
199 ret = ff_h2645_packet_split(&ctx->pkt, buf, buf_size, avctx, ctx->is_avc,
200 ctx->nal_length_size, AV_CODEC_ID_HEVC, 1, 0);
201 if (ret < 0)
202 return ret;
203
204 for (i = 0; i < ctx->pkt.nb_nals; i++) {
205 H2645NAL *nal = &ctx->pkt.nals[i];
206 GetBitContext *gb = &nal->gb;
207
208 if (nal->nuh_layer_id > 0)
209 continue;
210
211 switch (nal->type) {
212 case HEVC_NAL_VPS:
213 ff_hevc_decode_nal_vps(gb, avctx, ps);
214 break;
215 case HEVC_NAL_SPS:
216 ff_hevc_decode_nal_sps(gb, avctx, ps, 1);
217 break;
218 case HEVC_NAL_PPS:
219 ff_hevc_decode_nal_pps(gb, avctx, ps);
220 break;
221 case HEVC_NAL_SEI_PREFIX:
222 case HEVC_NAL_SEI_SUFFIX:
223 ff_hevc_decode_nal_sei(gb, avctx, sei, ps, nal->type);
224 break;
225 case HEVC_NAL_TRAIL_N:
226 case HEVC_NAL_TRAIL_R:
227 case HEVC_NAL_TSA_N:
228 case HEVC_NAL_TSA_R:
229 case HEVC_NAL_STSA_N:
230 case HEVC_NAL_STSA_R:
231 case HEVC_NAL_BLA_W_LP:
232 case HEVC_NAL_BLA_W_RADL:
233 case HEVC_NAL_BLA_N_LP:
234 case HEVC_NAL_IDR_W_RADL:
235 case HEVC_NAL_IDR_N_LP:
236 case HEVC_NAL_CRA_NUT:
237 case HEVC_NAL_RADL_N:
238 case HEVC_NAL_RADL_R:
239 case HEVC_NAL_RASL_N:
240 case HEVC_NAL_RASL_R:
241 if (ctx->sei.picture_timing.picture_struct == HEVC_SEI_PIC_STRUCT_FRAME_DOUBLING) {
242 s->repeat_pict = 1;
243 } else if (ctx->sei.picture_timing.picture_struct == HEVC_SEI_PIC_STRUCT_FRAME_TRIPLING) {
244 s->repeat_pict = 2;
245 }
246 ret = hevc_parse_slice_header(s, nal, avctx);
247 if (ret)
248 return ret;
249 break;
250 }
251 }
252 /* didn't find a picture! */
253 av_log(avctx, AV_LOG_ERROR, "missing picture in access unit with size %d\n", buf_size);
254 return -1;
255 }
256
257 /**
258 * Find the end of the current frame in the bitstream.
259 * @return the position of the first byte of the next frame, or END_NOT_FOUND
260 */
hevc_find_frame_end(AVCodecParserContext * s,const uint8_t * buf,int buf_size)261 static int hevc_find_frame_end(AVCodecParserContext *s, const uint8_t *buf,
262 int buf_size)
263 {
264 HEVCParserContext *ctx = s->priv_data;
265 ParseContext *pc = &ctx->pc;
266 int i;
267
268 for (i = 0; i < buf_size; i++) {
269 int nut;
270
271 pc->state64 = (pc->state64 << 8) | buf[i];
272
273 if (((pc->state64 >> 3 * 8) & 0xFFFFFF) != START_CODE)
274 continue;
275
276 nut = (pc->state64 >> 2 * 8 + 1) & 0x3F;
277 // Beginning of access unit
278 if ((nut >= HEVC_NAL_VPS && nut <= HEVC_NAL_EOB_NUT) || nut == HEVC_NAL_SEI_PREFIX ||
279 (nut >= 41 && nut <= 44) || (nut >= 48 && nut <= 55)) {
280 if (pc->frame_start_found) {
281 pc->frame_start_found = 0;
282 return i - 5;
283 }
284 } else if (nut <= HEVC_NAL_RASL_R ||
285 (nut >= HEVC_NAL_BLA_W_LP && nut <= HEVC_NAL_CRA_NUT)) {
286 int first_slice_segment_in_pic_flag = buf[i] >> 7;
287 if (first_slice_segment_in_pic_flag) {
288 if (!pc->frame_start_found) {
289 pc->frame_start_found = 1;
290 } else { // First slice of next frame found
291 pc->frame_start_found = 0;
292 return i - 5;
293 }
294 }
295 }
296 }
297
298 return END_NOT_FOUND;
299 }
300
hevc_parse(AVCodecParserContext * s,AVCodecContext * avctx,const uint8_t ** poutbuf,int * poutbuf_size,const uint8_t * buf,int buf_size)301 static int hevc_parse(AVCodecParserContext *s, AVCodecContext *avctx,
302 const uint8_t **poutbuf, int *poutbuf_size,
303 const uint8_t *buf, int buf_size)
304 {
305 int next;
306 HEVCParserContext *ctx = s->priv_data;
307 ParseContext *pc = &ctx->pc;
308 int is_dummy_buf = !buf_size;
309 const uint8_t *dummy_buf = buf;
310
311 if (avctx->extradata && !ctx->parsed_extradata) {
312 ff_hevc_decode_extradata(avctx->extradata, avctx->extradata_size, &ctx->ps, &ctx->sei,
313 &ctx->is_avc, &ctx->nal_length_size, avctx->err_recognition,
314 1, avctx);
315 ctx->parsed_extradata = 1;
316 }
317
318 if (s->flags & PARSER_FLAG_COMPLETE_FRAMES) {
319 next = buf_size;
320 } else {
321 next = hevc_find_frame_end(s, buf, buf_size);
322 if (ff_combine_frame(pc, next, &buf, &buf_size) < 0) {
323 *poutbuf = NULL;
324 *poutbuf_size = 0;
325 return buf_size;
326 }
327 }
328
329 is_dummy_buf &= (dummy_buf == buf);
330
331 if (!is_dummy_buf)
332 parse_nal_units(s, buf, buf_size, avctx);
333
334 *poutbuf = buf;
335 *poutbuf_size = buf_size;
336 return next;
337 }
338
339 // Split after the parameter sets at the beginning of the stream if they exist.
hevc_split(AVCodecContext * avctx,const uint8_t * buf,int buf_size)340 static int hevc_split(AVCodecContext *avctx, const uint8_t *buf, int buf_size)
341 {
342 const uint8_t *ptr = buf, *end = buf + buf_size;
343 uint32_t state = -1;
344 int has_vps = 0;
345 int has_sps = 0;
346 int has_pps = 0;
347 int nut;
348
349 while (ptr < end) {
350 ptr = avpriv_find_start_code(ptr, end, &state);
351 if ((state >> 8) != START_CODE)
352 break;
353 nut = (state >> 1) & 0x3F;
354 if (nut == HEVC_NAL_VPS)
355 has_vps = 1;
356 else if (nut == HEVC_NAL_SPS)
357 has_sps = 1;
358 else if (nut == HEVC_NAL_PPS)
359 has_pps = 1;
360 else if ((nut != HEVC_NAL_SEI_PREFIX || has_pps) &&
361 nut != HEVC_NAL_AUD) {
362 if (has_vps && has_sps) {
363 while (ptr - 4 > buf && ptr[-5] == 0)
364 ptr--;
365 return ptr - 4 - buf;
366 }
367 }
368 }
369 return 0;
370 }
371
hevc_parser_close(AVCodecParserContext * s)372 static void hevc_parser_close(AVCodecParserContext *s)
373 {
374 HEVCParserContext *ctx = s->priv_data;
375
376 ff_hevc_ps_uninit(&ctx->ps);
377 ff_h2645_packet_uninit(&ctx->pkt);
378 ff_hevc_reset_sei(&ctx->sei);
379
380 av_freep(&ctx->pc.buffer);
381 }
382
383 AVCodecParser ff_hevc_parser = {
384 .codec_ids = { AV_CODEC_ID_HEVC },
385 .priv_data_size = sizeof(HEVCParserContext),
386 .parser_parse = hevc_parse,
387 .parser_close = hevc_parser_close,
388 .split = hevc_split,
389 };
390