lavc: Add coded bitstream read/write support for H.264
[libav.git] / libavcodec / cbs_h264.h
1 /*
2 * This file is part of Libav.
3 *
4 * Libav is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
8 *
9 * Libav is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
13 *
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with Libav; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18
19 #ifndef AVCODEC_CBS_H264_H
20 #define AVCODEC_CBS_H264_H
21
22 #include <stddef.h>
23 #include <stdint.h>
24
25 #include "cbs_h2645.h"
26 #include "h264.h"
27
28
29 enum {
30 // This limit is arbitrary - it is sufficient for one message of each
31 // type plus some repeats, and will therefore easily cover all sane
32 // streams. However, it is possible to make technically-valid streams
33 // for which it will fail (for example, by including a large number of
34 // user-data-unregistered messages).
35 H264_MAX_SEI_PAYLOADS = 64,
36 };
37
38
39 typedef struct H264RawNALUnitHeader {
40 uint8_t forbidden_zero_bit;
41 uint8_t nal_ref_idc;
42 uint8_t nal_unit_type;
43
44 uint8_t svc_extension_flag;
45 uint8_t avc_3d_extension_flag;
46 } H264RawNALUnitHeader;
47
48 typedef struct H264RawScalingList {
49 int8_t delta_scale[64];
50 } H264RawScalingList;
51
52 typedef struct H264RawHRD {
53 uint8_t cpb_cnt_minus1;
54 uint8_t bit_rate_scale;
55 uint8_t cpb_size_scale;
56
57 uint32_t bit_rate_value_minus1[H264_MAX_CPB_CNT];
58 uint32_t cpb_size_value_minus1[H264_MAX_CPB_CNT];
59 uint8_t cbr_flag[H264_MAX_CPB_CNT];
60
61 uint8_t initial_cpb_removal_delay_length_minus1;
62 uint8_t cpb_removal_delay_length_minus1;
63 uint8_t dpb_output_delay_length_minus1;
64 uint8_t time_offset_length;
65 } H264RawHRD;
66
67 typedef struct H264RawVUI {
68 uint8_t aspect_ratio_info_present_flag;
69 uint8_t aspect_ratio_idc;
70 uint16_t sar_width;
71 uint16_t sar_height;
72
73 uint8_t overscan_info_present_flag;
74 uint8_t overscan_appropriate_flag;
75
76 uint8_t video_signal_type_present_flag;
77 uint8_t video_format;
78 uint8_t video_full_range_flag;
79 uint8_t colour_description_present_flag;
80 uint8_t colour_primaries;
81 uint8_t transfer_characteristics;
82 uint8_t matrix_coefficients;
83
84 uint8_t chroma_loc_info_present_flag;
85 uint8_t chroma_sample_loc_type_top_field;
86 uint8_t chroma_sample_loc_type_bottom_field;
87
88 uint8_t timing_info_present_flag;
89 uint32_t num_units_in_tick;
90 uint32_t time_scale;
91 uint8_t fixed_frame_rate_flag;
92
93 uint8_t nal_hrd_parameters_present_flag;
94 H264RawHRD nal_hrd_parameters;
95 uint8_t vcl_hrd_parameters_present_flag;
96 H264RawHRD vcl_hrd_parameters;
97 uint8_t low_delay_hrd_flag;
98
99 uint8_t pic_struct_present_flag;
100
101 uint8_t bitstream_restriction_flag;
102 uint8_t motion_vectors_over_pic_boundaries_flag;
103 uint8_t max_bytes_per_pic_denom;
104 uint8_t max_bits_per_mb_denom;
105 uint8_t log2_max_mv_length_horizontal;
106 uint8_t log2_max_mv_length_vertical;
107 uint8_t max_num_reorder_frames;
108 uint8_t max_dec_frame_buffering;
109 } H264RawVUI;
110
111 typedef struct H264RawSPS {
112 H264RawNALUnitHeader nal_unit_header;
113
114 uint8_t profile_idc;
115 uint8_t constraint_set0_flag;
116 uint8_t constraint_set1_flag;
117 uint8_t constraint_set2_flag;
118 uint8_t constraint_set3_flag;
119 uint8_t constraint_set4_flag;
120 uint8_t constraint_set5_flag;
121 uint8_t reserved_zero_2bits;
122 uint8_t level_idc;
123
124 uint8_t seq_parameter_set_id;
125
126 uint8_t chroma_format_idc;
127 uint8_t separate_colour_plane_flag;
128 uint8_t bit_depth_luma_minus8;
129 uint8_t bit_depth_chroma_minus8;
130 uint8_t qpprime_y_zero_transform_bypass_flag;
131
132 uint8_t seq_scaling_matrix_present_flag;
133 uint8_t seq_scaling_list_present_flag[12];
134 H264RawScalingList scaling_list_4x4[6];
135 H264RawScalingList scaling_list_8x8[6];
136
137 uint8_t log2_max_frame_num_minus4;
138 uint8_t pic_order_cnt_type;
139 uint8_t log2_max_pic_order_cnt_lsb_minus4;
140 uint8_t delta_pic_order_always_zero_flag;
141 int32_t offset_for_non_ref_pic;
142 int32_t offset_for_top_to_bottom_field;
143 uint8_t num_ref_frames_in_pic_order_cnt_cycle;
144 int32_t offset_for_ref_frame[256];
145
146 uint8_t max_num_ref_frames;
147 uint8_t gaps_in_frame_num_allowed_flag;
148
149 uint16_t pic_width_in_mbs_minus1;
150 uint16_t pic_height_in_map_units_minus1;
151
152 uint8_t frame_mbs_only_flag;
153 uint8_t mb_adaptive_frame_field_flag;
154 uint8_t direct_8x8_inference_flag;
155
156 uint8_t frame_cropping_flag;
157 uint16_t frame_crop_left_offset;
158 uint16_t frame_crop_right_offset;
159 uint16_t frame_crop_top_offset;
160 uint16_t frame_crop_bottom_offset;
161
162 uint8_t vui_parameters_present_flag;
163 H264RawVUI vui;
164 } H264RawSPS;
165
166 typedef struct H264RawSPSExtension {
167 H264RawNALUnitHeader nal_unit_header;
168
169 uint8_t seq_parameter_set_id;
170
171 uint8_t aux_format_idc;
172 uint8_t bit_depth_aux_minus8;
173 uint8_t alpha_incr_flag;
174 uint16_t alpha_opaque_value;
175 uint16_t alpha_transparent_value;
176
177 uint8_t additional_extension_flag;
178 } H264RawSPSExtension;
179
180 typedef struct H264RawPPS {
181 H264RawNALUnitHeader nal_unit_header;
182
183 uint8_t pic_parameter_set_id;
184 uint8_t seq_parameter_set_id;
185
186 uint8_t entropy_coding_mode_flag;
187 uint8_t bottom_field_pic_order_in_frame_present_flag;
188
189 uint8_t num_slice_groups_minus1;
190 uint8_t slice_group_map_type;
191 uint16_t run_length_minus1[H264_MAX_SLICE_GROUPS];
192 uint16_t top_left[H264_MAX_SLICE_GROUPS];
193 uint16_t bottom_right[H264_MAX_SLICE_GROUPS];
194 uint8_t slice_group_change_direction_flag;
195 uint16_t slice_group_change_rate_minus1;
196 uint16_t pic_size_in_map_units_minus1;
197 uint8_t slice_group_id[H264_MAX_MB_PIC_SIZE];
198
199 uint8_t num_ref_idx_l0_default_active_minus1;
200 uint8_t num_ref_idx_l1_default_active_minus1;
201
202 uint8_t weighted_pred_flag;
203 uint8_t weighted_bipred_idc;
204
205 int8_t pic_init_qp_minus26;
206 int8_t pic_init_qs_minus26;
207 int8_t chroma_qp_index_offset;
208
209 uint8_t deblocking_filter_control_present_flag;
210 uint8_t constrained_intra_pred_flag;
211
212 uint8_t more_rbsp_data;
213
214 uint8_t redundant_pic_cnt_present_flag;
215 uint8_t transform_8x8_mode_flag;
216
217 uint8_t pic_scaling_matrix_present_flag;
218 uint8_t pic_scaling_list_present_flag[12];
219 H264RawScalingList scaling_list_4x4[6];
220 H264RawScalingList scaling_list_8x8[6];
221
222 int8_t second_chroma_qp_index_offset;
223 } H264RawPPS;
224
225 typedef struct H264RawAUD {
226 H264RawNALUnitHeader nal_unit_header;
227
228 uint8_t primary_pic_type;
229 } H264RawAUD;
230
231 typedef struct H264RawSEIBufferingPeriod {
232 uint8_t seq_parameter_set_id;
233 struct {
234 uint32_t initial_cpb_removal_delay[H264_MAX_CPB_CNT];
235 uint32_t initial_cpb_removal_delay_offset[H264_MAX_CPB_CNT];
236 } nal, vcl;
237 } H264RawSEIBufferingPeriod;
238
239 typedef struct H264RawSEIPicTimestamp {
240 uint8_t ct_type;
241 uint8_t nuit_field_based_flag;
242 uint8_t counting_type;
243 uint8_t full_timestamp_flag;
244 uint8_t discontinuity_flag;
245 uint8_t cnt_dropped_flag;
246 uint8_t n_frames;
247 uint8_t seconds_flag;
248 uint8_t seconds_value;
249 uint8_t minutes_flag;
250 uint8_t minutes_value;
251 uint8_t hours_flag;
252 uint8_t hours_value;
253 uint32_t time_offset;
254 } H264RawSEIPicTimestamp;
255
256 typedef struct H264RawSEIPicTiming {
257 uint32_t cpb_removal_delay;
258 uint32_t dpb_output_delay;
259 uint8_t pic_struct;
260 uint8_t clock_timestamp_flag[3];
261 H264RawSEIPicTimestamp timestamp[3];
262 } H264RawSEIPicTiming;
263
264 typedef struct H264RawSEIUserDataRegistered {
265 uint8_t itu_t_t35_country_code;
266 uint8_t itu_t_t35_country_code_extension_byte;
267 uint8_t *data;
268 size_t data_length;
269 } H264RawSEIUserDataRegistered;
270
271 typedef struct H264RawSEIUserDataUnregistered {
272 uint8_t uuid_iso_iec_11578[16];
273 uint8_t *data;
274 size_t data_length;
275 } H264RawSEIUserDataUnregistered;
276
277 typedef struct H264RawSEIRecoveryPoint {
278 uint16_t recovery_frame_cnt;
279 uint8_t exact_match_flag;
280 uint8_t broken_link_flag;
281 uint8_t changing_slice_group_idc;
282 } H264RawSEIRecoveryPoint;
283
284 typedef struct H264RawSEIDisplayOrientation {
285 uint8_t display_orientation_cancel_flag;
286 uint8_t hor_flip;
287 uint8_t ver_flip;
288 uint16_t anticlockwise_rotation;
289 uint16_t display_orientation_repetition_period;
290 uint8_t display_orientation_extension_flag;
291 } H264RawSEIDisplayOrientation;
292
293 typedef struct H264RawSEIPayload {
294 uint32_t payload_type;
295 uint32_t payload_size;
296 union {
297 H264RawSEIBufferingPeriod buffering_period;
298 H264RawSEIPicTiming pic_timing;
299 // H264RawSEIFiller filler -> no fields.
300 H264RawSEIUserDataRegistered user_data_registered;
301 H264RawSEIUserDataUnregistered user_data_unregistered;
302 H264RawSEIRecoveryPoint recovery_point;
303 H264RawSEIDisplayOrientation display_orientation;
304 struct {
305 uint8_t *data;
306 size_t data_length;
307 } other;
308 } payload;
309 } H264RawSEIPayload;
310
311 typedef struct H264RawSEI {
312 H264RawNALUnitHeader nal_unit_header;
313
314 H264RawSEIPayload payload[H264_MAX_SEI_PAYLOADS];
315 uint8_t payload_count;
316 } H264RawSEI;
317
318 typedef struct H264RawSliceHeader {
319 H264RawNALUnitHeader nal_unit_header;
320
321 uint32_t first_mb_in_slice;
322 uint8_t slice_type;
323
324 uint8_t pic_parameter_set_id;
325
326 uint8_t colour_plane_id;
327
328 uint16_t frame_num;
329 uint8_t field_pic_flag;
330 uint8_t bottom_field_flag;
331
332 uint16_t idr_pic_id;
333
334 uint16_t pic_order_cnt_lsb;
335 int32_t delta_pic_order_cnt_bottom;
336 int32_t delta_pic_order_cnt[2];
337
338 uint8_t redundant_pic_cnt;
339 uint8_t direct_spatial_mv_pred_flag;
340
341 uint8_t num_ref_idx_active_override_flag;
342 uint8_t num_ref_idx_l0_active_minus1;
343 uint8_t num_ref_idx_l1_active_minus1;
344
345 uint8_t ref_pic_list_modification_flag_l0;
346 uint8_t ref_pic_list_modification_flag_l1;
347 struct {
348 uint8_t modification_of_pic_nums_idc;
349 int32_t abs_diff_pic_num_minus1;
350 uint8_t long_term_pic_num;
351 } rplm_l0[H264_MAX_RPLM_COUNT], rplm_l1[H264_MAX_RPLM_COUNT];
352
353 uint8_t luma_log2_weight_denom;
354 uint8_t chroma_log2_weight_denom;
355
356 uint8_t luma_weight_l0_flag[H264_MAX_REFS];
357 int8_t luma_weight_l0[H264_MAX_REFS];
358 int8_t luma_offset_l0[H264_MAX_REFS];
359 uint8_t chroma_weight_l0_flag[H264_MAX_REFS];
360 int8_t chroma_weight_l0[H264_MAX_REFS][2];
361 int8_t chroma_offset_l0[H264_MAX_REFS][2];
362
363 uint8_t luma_weight_l1_flag[H264_MAX_REFS];
364 int8_t luma_weight_l1[H264_MAX_REFS];
365 int8_t luma_offset_l1[H264_MAX_REFS];
366 uint8_t chroma_weight_l1_flag[H264_MAX_REFS];
367 int8_t chroma_weight_l1[H264_MAX_REFS][2];
368 int8_t chroma_offset_l1[H264_MAX_REFS][2];
369
370 uint8_t no_output_of_prior_pics_flag;
371 uint8_t long_term_reference_flag;
372
373 uint8_t adaptive_ref_pic_marking_mode_flag;
374 struct {
375 uint8_t memory_management_control_operation;
376 int32_t difference_of_pic_nums_minus1;
377 uint8_t long_term_pic_num;
378 uint8_t long_term_frame_idx;
379 uint8_t max_long_term_frame_idx_plus1;
380 } mmco[H264_MAX_MMCO_COUNT];
381
382 uint8_t cabac_init_idc;
383
384 int8_t slice_qp_delta;
385
386 uint8_t sp_for_switch_flag;
387 int8_t slice_qs_delta;
388
389 uint8_t disable_deblocking_filter_idc;
390 int8_t slice_alpha_c0_offset_div2;
391 int8_t slice_beta_offset_div2;
392
393 uint16_t slice_group_change_cycle;
394 } H264RawSliceHeader;
395
396 typedef struct H264RawSlice {
397 H264RawSliceHeader header;
398
399 uint8_t *data;
400 size_t data_size;
401 int data_bit_start;
402 } H264RawSlice;
403
404
405 typedef struct CodedBitstreamH264Context {
406 // Reader/writer context in common with the H.265 implementation.
407 CodedBitstreamH2645Context common;
408
409 // All currently available parameter sets. These are updated when
410 // any parameter set NAL unit is read/written with this context.
411 H264RawSPS *sps[H264_MAX_SPS_COUNT];
412 H264RawPPS *pps[H264_MAX_PPS_COUNT];
413
414 // The currently active parameter sets. These are updated when any
415 // NAL unit refers to the relevant parameter set. These pointers
416 // must also be present in the arrays above.
417 const H264RawSPS *active_sps;
418 const H264RawPPS *active_pps;
419
420 // The NAL unit type of the most recent normal slice. This is required
421 // to be able to read/write auxiliary slices, because IdrPicFlag is
422 // otherwise unknown.
423 uint8_t last_slice_nal_unit_type;
424 } CodedBitstreamH264Context;
425
426
427 #endif /* AVCODEC_CBS_H264_H */