You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

459 lines
13KB

  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. #ifndef AVCODEC_CBS_H264_H
  19. #define AVCODEC_CBS_H264_H
  20. #include <stddef.h>
  21. #include <stdint.h>
  22. #include "cbs.h"
  23. #include "cbs_h2645.h"
  24. #include "h264.h"
  25. enum {
  26. // This limit is arbitrary - it is sufficient for one message of each
  27. // type plus some repeats, and will therefore easily cover all sane
  28. // streams. However, it is possible to make technically-valid streams
  29. // for which it will fail (for example, by including a large number of
  30. // user-data-unregistered messages).
  31. H264_MAX_SEI_PAYLOADS = 64,
  32. };
  33. typedef struct H264RawNALUnitHeader {
  34. uint8_t forbidden_zero_bit;
  35. uint8_t nal_ref_idc;
  36. uint8_t nal_unit_type;
  37. uint8_t svc_extension_flag;
  38. uint8_t avc_3d_extension_flag;
  39. } H264RawNALUnitHeader;
  40. typedef struct H264RawScalingList {
  41. int8_t delta_scale[64];
  42. } H264RawScalingList;
  43. typedef struct H264RawHRD {
  44. uint8_t cpb_cnt_minus1;
  45. uint8_t bit_rate_scale;
  46. uint8_t cpb_size_scale;
  47. uint32_t bit_rate_value_minus1[H264_MAX_CPB_CNT];
  48. uint32_t cpb_size_value_minus1[H264_MAX_CPB_CNT];
  49. uint8_t cbr_flag[H264_MAX_CPB_CNT];
  50. uint8_t initial_cpb_removal_delay_length_minus1;
  51. uint8_t cpb_removal_delay_length_minus1;
  52. uint8_t dpb_output_delay_length_minus1;
  53. uint8_t time_offset_length;
  54. } H264RawHRD;
  55. typedef struct H264RawVUI {
  56. uint8_t aspect_ratio_info_present_flag;
  57. uint8_t aspect_ratio_idc;
  58. uint16_t sar_width;
  59. uint16_t sar_height;
  60. uint8_t overscan_info_present_flag;
  61. uint8_t overscan_appropriate_flag;
  62. uint8_t video_signal_type_present_flag;
  63. uint8_t video_format;
  64. uint8_t video_full_range_flag;
  65. uint8_t colour_description_present_flag;
  66. uint8_t colour_primaries;
  67. uint8_t transfer_characteristics;
  68. uint8_t matrix_coefficients;
  69. uint8_t chroma_loc_info_present_flag;
  70. uint8_t chroma_sample_loc_type_top_field;
  71. uint8_t chroma_sample_loc_type_bottom_field;
  72. uint8_t timing_info_present_flag;
  73. uint32_t num_units_in_tick;
  74. uint32_t time_scale;
  75. uint8_t fixed_frame_rate_flag;
  76. uint8_t nal_hrd_parameters_present_flag;
  77. H264RawHRD nal_hrd_parameters;
  78. uint8_t vcl_hrd_parameters_present_flag;
  79. H264RawHRD vcl_hrd_parameters;
  80. uint8_t low_delay_hrd_flag;
  81. uint8_t pic_struct_present_flag;
  82. uint8_t bitstream_restriction_flag;
  83. uint8_t motion_vectors_over_pic_boundaries_flag;
  84. uint8_t max_bytes_per_pic_denom;
  85. uint8_t max_bits_per_mb_denom;
  86. uint8_t log2_max_mv_length_horizontal;
  87. uint8_t log2_max_mv_length_vertical;
  88. uint8_t max_num_reorder_frames;
  89. uint8_t max_dec_frame_buffering;
  90. } H264RawVUI;
  91. typedef struct H264RawSPS {
  92. H264RawNALUnitHeader nal_unit_header;
  93. uint8_t profile_idc;
  94. uint8_t constraint_set0_flag;
  95. uint8_t constraint_set1_flag;
  96. uint8_t constraint_set2_flag;
  97. uint8_t constraint_set3_flag;
  98. uint8_t constraint_set4_flag;
  99. uint8_t constraint_set5_flag;
  100. uint8_t reserved_zero_2bits;
  101. uint8_t level_idc;
  102. uint8_t seq_parameter_set_id;
  103. uint8_t chroma_format_idc;
  104. uint8_t separate_colour_plane_flag;
  105. uint8_t bit_depth_luma_minus8;
  106. uint8_t bit_depth_chroma_minus8;
  107. uint8_t qpprime_y_zero_transform_bypass_flag;
  108. uint8_t seq_scaling_matrix_present_flag;
  109. uint8_t seq_scaling_list_present_flag[12];
  110. H264RawScalingList scaling_list_4x4[6];
  111. H264RawScalingList scaling_list_8x8[6];
  112. uint8_t log2_max_frame_num_minus4;
  113. uint8_t pic_order_cnt_type;
  114. uint8_t log2_max_pic_order_cnt_lsb_minus4;
  115. uint8_t delta_pic_order_always_zero_flag;
  116. int32_t offset_for_non_ref_pic;
  117. int32_t offset_for_top_to_bottom_field;
  118. uint8_t num_ref_frames_in_pic_order_cnt_cycle;
  119. int32_t offset_for_ref_frame[256];
  120. uint8_t max_num_ref_frames;
  121. uint8_t gaps_in_frame_num_allowed_flag;
  122. uint16_t pic_width_in_mbs_minus1;
  123. uint16_t pic_height_in_map_units_minus1;
  124. uint8_t frame_mbs_only_flag;
  125. uint8_t mb_adaptive_frame_field_flag;
  126. uint8_t direct_8x8_inference_flag;
  127. uint8_t frame_cropping_flag;
  128. uint16_t frame_crop_left_offset;
  129. uint16_t frame_crop_right_offset;
  130. uint16_t frame_crop_top_offset;
  131. uint16_t frame_crop_bottom_offset;
  132. uint8_t vui_parameters_present_flag;
  133. H264RawVUI vui;
  134. } H264RawSPS;
  135. typedef struct H264RawSPSExtension {
  136. H264RawNALUnitHeader nal_unit_header;
  137. uint8_t seq_parameter_set_id;
  138. uint8_t aux_format_idc;
  139. uint8_t bit_depth_aux_minus8;
  140. uint8_t alpha_incr_flag;
  141. uint16_t alpha_opaque_value;
  142. uint16_t alpha_transparent_value;
  143. uint8_t additional_extension_flag;
  144. } H264RawSPSExtension;
  145. typedef struct H264RawPPS {
  146. H264RawNALUnitHeader nal_unit_header;
  147. uint8_t pic_parameter_set_id;
  148. uint8_t seq_parameter_set_id;
  149. uint8_t entropy_coding_mode_flag;
  150. uint8_t bottom_field_pic_order_in_frame_present_flag;
  151. uint8_t num_slice_groups_minus1;
  152. uint8_t slice_group_map_type;
  153. uint16_t run_length_minus1[H264_MAX_SLICE_GROUPS];
  154. uint16_t top_left[H264_MAX_SLICE_GROUPS];
  155. uint16_t bottom_right[H264_MAX_SLICE_GROUPS];
  156. uint8_t slice_group_change_direction_flag;
  157. uint16_t slice_group_change_rate_minus1;
  158. uint16_t pic_size_in_map_units_minus1;
  159. uint8_t *slice_group_id;
  160. AVBufferRef *slice_group_id_ref;
  161. uint8_t num_ref_idx_l0_default_active_minus1;
  162. uint8_t num_ref_idx_l1_default_active_minus1;
  163. uint8_t weighted_pred_flag;
  164. uint8_t weighted_bipred_idc;
  165. int8_t pic_init_qp_minus26;
  166. int8_t pic_init_qs_minus26;
  167. int8_t chroma_qp_index_offset;
  168. uint8_t deblocking_filter_control_present_flag;
  169. uint8_t constrained_intra_pred_flag;
  170. uint8_t more_rbsp_data;
  171. uint8_t redundant_pic_cnt_present_flag;
  172. uint8_t transform_8x8_mode_flag;
  173. uint8_t pic_scaling_matrix_present_flag;
  174. uint8_t pic_scaling_list_present_flag[12];
  175. H264RawScalingList scaling_list_4x4[6];
  176. H264RawScalingList scaling_list_8x8[6];
  177. int8_t second_chroma_qp_index_offset;
  178. } H264RawPPS;
  179. typedef struct H264RawAUD {
  180. H264RawNALUnitHeader nal_unit_header;
  181. uint8_t primary_pic_type;
  182. } H264RawAUD;
  183. typedef struct H264RawSEIBufferingPeriod {
  184. uint8_t seq_parameter_set_id;
  185. struct {
  186. uint32_t initial_cpb_removal_delay[H264_MAX_CPB_CNT];
  187. uint32_t initial_cpb_removal_delay_offset[H264_MAX_CPB_CNT];
  188. } nal, vcl;
  189. } H264RawSEIBufferingPeriod;
  190. typedef struct H264RawSEIPicTimestamp {
  191. uint8_t ct_type;
  192. uint8_t nuit_field_based_flag;
  193. uint8_t counting_type;
  194. uint8_t full_timestamp_flag;
  195. uint8_t discontinuity_flag;
  196. uint8_t cnt_dropped_flag;
  197. uint8_t n_frames;
  198. uint8_t seconds_flag;
  199. uint8_t seconds_value;
  200. uint8_t minutes_flag;
  201. uint8_t minutes_value;
  202. uint8_t hours_flag;
  203. uint8_t hours_value;
  204. uint32_t time_offset;
  205. } H264RawSEIPicTimestamp;
  206. typedef struct H264RawSEIPicTiming {
  207. uint32_t cpb_removal_delay;
  208. uint32_t dpb_output_delay;
  209. uint8_t pic_struct;
  210. uint8_t clock_timestamp_flag[3];
  211. H264RawSEIPicTimestamp timestamp[3];
  212. } H264RawSEIPicTiming;
  213. typedef struct H264RawSEIUserDataRegistered {
  214. uint8_t itu_t_t35_country_code;
  215. uint8_t itu_t_t35_country_code_extension_byte;
  216. uint8_t *data;
  217. size_t data_length;
  218. AVBufferRef *data_ref;
  219. } H264RawSEIUserDataRegistered;
  220. typedef struct H264RawSEIUserDataUnregistered {
  221. uint8_t uuid_iso_iec_11578[16];
  222. uint8_t *data;
  223. size_t data_length;
  224. AVBufferRef *data_ref;
  225. } H264RawSEIUserDataUnregistered;
  226. typedef struct H264RawSEIRecoveryPoint {
  227. uint16_t recovery_frame_cnt;
  228. uint8_t exact_match_flag;
  229. uint8_t broken_link_flag;
  230. uint8_t changing_slice_group_idc;
  231. } H264RawSEIRecoveryPoint;
  232. typedef struct H264RawSEIDisplayOrientation {
  233. uint8_t display_orientation_cancel_flag;
  234. uint8_t hor_flip;
  235. uint8_t ver_flip;
  236. uint16_t anticlockwise_rotation;
  237. uint16_t display_orientation_repetition_period;
  238. uint8_t display_orientation_extension_flag;
  239. } H264RawSEIDisplayOrientation;
  240. typedef struct H264RawSEIPayload {
  241. uint32_t payload_type;
  242. uint32_t payload_size;
  243. union {
  244. H264RawSEIBufferingPeriod buffering_period;
  245. H264RawSEIPicTiming pic_timing;
  246. // H264RawSEIFiller filler -> no fields.
  247. H264RawSEIUserDataRegistered user_data_registered;
  248. H264RawSEIUserDataUnregistered user_data_unregistered;
  249. H264RawSEIRecoveryPoint recovery_point;
  250. H264RawSEIDisplayOrientation display_orientation;
  251. struct {
  252. uint8_t *data;
  253. size_t data_length;
  254. AVBufferRef *data_ref;
  255. } other;
  256. } payload;
  257. } H264RawSEIPayload;
  258. typedef struct H264RawSEI {
  259. H264RawNALUnitHeader nal_unit_header;
  260. H264RawSEIPayload payload[H264_MAX_SEI_PAYLOADS];
  261. uint8_t payload_count;
  262. } H264RawSEI;
  263. typedef struct H264RawSliceHeader {
  264. H264RawNALUnitHeader nal_unit_header;
  265. uint32_t first_mb_in_slice;
  266. uint8_t slice_type;
  267. uint8_t pic_parameter_set_id;
  268. uint8_t colour_plane_id;
  269. uint16_t frame_num;
  270. uint8_t field_pic_flag;
  271. uint8_t bottom_field_flag;
  272. uint16_t idr_pic_id;
  273. uint16_t pic_order_cnt_lsb;
  274. int32_t delta_pic_order_cnt_bottom;
  275. int32_t delta_pic_order_cnt[2];
  276. uint8_t redundant_pic_cnt;
  277. uint8_t direct_spatial_mv_pred_flag;
  278. uint8_t num_ref_idx_active_override_flag;
  279. uint8_t num_ref_idx_l0_active_minus1;
  280. uint8_t num_ref_idx_l1_active_minus1;
  281. uint8_t ref_pic_list_modification_flag_l0;
  282. uint8_t ref_pic_list_modification_flag_l1;
  283. struct {
  284. uint8_t modification_of_pic_nums_idc;
  285. int32_t abs_diff_pic_num_minus1;
  286. uint8_t long_term_pic_num;
  287. } rplm_l0[H264_MAX_RPLM_COUNT], rplm_l1[H264_MAX_RPLM_COUNT];
  288. uint8_t luma_log2_weight_denom;
  289. uint8_t chroma_log2_weight_denom;
  290. uint8_t luma_weight_l0_flag[H264_MAX_REFS];
  291. int8_t luma_weight_l0[H264_MAX_REFS];
  292. int8_t luma_offset_l0[H264_MAX_REFS];
  293. uint8_t chroma_weight_l0_flag[H264_MAX_REFS];
  294. int8_t chroma_weight_l0[H264_MAX_REFS][2];
  295. int8_t chroma_offset_l0[H264_MAX_REFS][2];
  296. uint8_t luma_weight_l1_flag[H264_MAX_REFS];
  297. int8_t luma_weight_l1[H264_MAX_REFS];
  298. int8_t luma_offset_l1[H264_MAX_REFS];
  299. uint8_t chroma_weight_l1_flag[H264_MAX_REFS];
  300. int8_t chroma_weight_l1[H264_MAX_REFS][2];
  301. int8_t chroma_offset_l1[H264_MAX_REFS][2];
  302. uint8_t no_output_of_prior_pics_flag;
  303. uint8_t long_term_reference_flag;
  304. uint8_t adaptive_ref_pic_marking_mode_flag;
  305. struct {
  306. uint8_t memory_management_control_operation;
  307. int32_t difference_of_pic_nums_minus1;
  308. uint8_t long_term_pic_num;
  309. uint8_t long_term_frame_idx;
  310. uint8_t max_long_term_frame_idx_plus1;
  311. } mmco[H264_MAX_MMCO_COUNT];
  312. uint8_t cabac_init_idc;
  313. int8_t slice_qp_delta;
  314. uint8_t sp_for_switch_flag;
  315. int8_t slice_qs_delta;
  316. uint8_t disable_deblocking_filter_idc;
  317. int8_t slice_alpha_c0_offset_div2;
  318. int8_t slice_beta_offset_div2;
  319. uint16_t slice_group_change_cycle;
  320. } H264RawSliceHeader;
  321. typedef struct H264RawSlice {
  322. H264RawSliceHeader header;
  323. uint8_t *data;
  324. size_t data_size;
  325. int data_bit_start;
  326. AVBufferRef *data_ref;
  327. } H264RawSlice;
  328. typedef struct H264RawFiller {
  329. H264RawNALUnitHeader nal_unit_header;
  330. uint32_t filler_size;
  331. } H264RawFiller;
  332. typedef struct CodedBitstreamH264Context {
  333. // Reader/writer context in common with the H.265 implementation.
  334. CodedBitstreamH2645Context common;
  335. // All currently available parameter sets. These are updated when
  336. // any parameter set NAL unit is read/written with this context.
  337. H264RawSPS *sps[H264_MAX_SPS_COUNT];
  338. H264RawPPS *pps[H264_MAX_PPS_COUNT];
  339. // The currently active parameter sets. These are updated when any
  340. // NAL unit refers to the relevant parameter set. These pointers
  341. // must also be present in the arrays above.
  342. const H264RawSPS *active_sps;
  343. const H264RawPPS *active_pps;
  344. // The NAL unit type of the most recent normal slice. This is required
  345. // to be able to read/write auxiliary slices, because IdrPicFlag is
  346. // otherwise unknown.
  347. uint8_t last_slice_nal_unit_type;
  348. } CodedBitstreamH264Context;
  349. /**
  350. * Add an SEI message to an access unit.
  351. */
  352. int ff_cbs_h264_add_sei_message(CodedBitstreamContext *ctx,
  353. CodedBitstreamFragment *access_unit,
  354. const H264RawSEIPayload *payload);
  355. /**
  356. * Delete an SEI message from an access unit.
  357. *
  358. * Deletes from nal_unit, which must be an SEI NAL unit. If this is the
  359. * last message in nal_unit, also deletes it from access_unit.
  360. */
  361. int ff_cbs_h264_delete_sei_message(CodedBitstreamContext *ctx,
  362. CodedBitstreamFragment *access_unit,
  363. CodedBitstreamUnit *nal_unit,
  364. int position);
  365. #endif /* AVCODEC_CBS_H264_H */