libavcodec/h264_sei.c

   1 /*
   2  * H.26L/H.264/AVC/JVT/14496-10/... SEI decoding
   3  * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav 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  * Libav 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 Libav; 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 SEI decoding.
  25  * @author Michael Niedermayer <michaelni@gmx.at>
  26  */
  27
  28 #include "avcodec.h"
  29 #include "get_bits.h"
  30 #include "golomb_legacy.h"
  31 #include "h264_ps.h"
  32 #include "h264_sei.h"
  33 #include "internal.h"
  34
  35 static const uint8_t sei_num_clock_ts_table[9] = {
  36     1, 1, 1, 2, 2, 3, 3, 2, 3
  37 };
  38
  39 void ff_h264_sei_uninit(H264SEIContext *h)
  40 {
  41     h->unregistered.x264_build           = -1;
  42     h->recovery_point.recovery_frame_cnt = -1;
  43
  44     h->picture_timing.dpb_output_delay  = 0;
  45     h->picture_timing.cpb_removal_delay = -1;
  46
  47     h->picture_timing.present      = 0;
  48     h->buffering_period.present    = 0;
  49     h->frame_packing.present       = 0;
  50     h->display_orientation.present = 0;
  51     h->afd.present                 =  0;
  52
  53     h->a53_caption.a53_caption_size = 0;
  54     av_freep(&h->a53_caption.a53_caption);
  55 }
  56
  57 static int decode_picture_timing(H264SEIPictureTiming *h, GetBitContext *gb,
  58                                  const SPS *sps, void *logctx)
  59 {
  60     if (!sps)
  61         return AVERROR_INVALIDDATA;
  62
  63     if (sps->nal_hrd_parameters_present_flag ||
  64         sps->vcl_hrd_parameters_present_flag) {
  65         h->cpb_removal_delay = get_bits(gb, sps->cpb_removal_delay_length);
  66         h->dpb_output_delay  = get_bits(gb, sps->dpb_output_delay_length);
  67     }
  68     if (sps->pic_struct_present_flag) {
  69         unsigned int i, num_clock_ts;
  70
  71         h->pic_struct = get_bits(gb, 4);
  72         h->ct_type    = 0;
  73
  74         if (h->pic_struct > H264_SEI_PIC_STRUCT_FRAME_TRIPLING)
  75             return AVERROR_INVALIDDATA;
  76
  77         num_clock_ts = sei_num_clock_ts_table[h->pic_struct];
  78
  79         for (i = 0; i < num_clock_ts; i++) {
  80             if (get_bits(gb, 1)) {                /* clock_timestamp_flag */
  81                 unsigned int full_timestamp_flag;
  82
  83                 h->ct_type |= 1 << get_bits(gb, 2);
  84                 skip_bits(gb, 1);                 /* nuit_field_based_flag */
  85                 skip_bits(gb, 5);                 /* counting_type */
  86                 full_timestamp_flag = get_bits(gb, 1);
  87                 skip_bits(gb, 1);                 /* discontinuity_flag */
  88                 skip_bits(gb, 1);                 /* cnt_dropped_flag */
  89                 skip_bits(gb, 8);                 /* n_frames */
  90                 if (full_timestamp_flag) {
  91                     skip_bits(gb, 6);             /* seconds_value 0..59 */
  92                     skip_bits(gb, 6);             /* minutes_value 0..59 */
  93                     skip_bits(gb, 5);             /* hours_value 0..23 */
  94                 } else {
  95                     if (get_bits(gb, 1)) {        /* seconds_flag */
  96                         skip_bits(gb, 6);         /* seconds_value range 0..59 */
  97                         if (get_bits(gb, 1)) {    /* minutes_flag */
  98                             skip_bits(gb, 6);     /* minutes_value 0..59 */
  99                             if (get_bits(gb, 1))  /* hours_flag */
 100                                 skip_bits(gb, 5); /* hours_value 0..23 */
 101                         }
 102                     }
 103                 }
 104                 if (sps->time_offset_length > 0)
 105                     skip_bits(gb,
 106                               sps->time_offset_length); /* time_offset */
 107             }
 108         }
 109
 110         av_log(logctx, AV_LOG_DEBUG, "ct_type:%X pic_struct:%d\n",
 111                h->ct_type, h->pic_struct);
 112     }
 113
 114     h->present = 1;
 115     return 0;
 116 }
 117
 118 static int decode_registered_user_data_afd(H264SEIAFD *h, GetBitContext *gb, int size)
 119 {
 120     int flag;
 121
 122     if (size-- < 1)
 123         return AVERROR_INVALIDDATA;
 124     skip_bits(gb, 1);               // 0
 125     flag = get_bits(gb, 1);         // active_format_flag
 126     skip_bits(gb, 6);               // reserved
 127
 128     if (flag) {
 129         if (size-- < 1)
 130             return AVERROR_INVALIDDATA;
 131         skip_bits(gb, 4);           // reserved
 132         h->active_format_description = get_bits(gb, 4);
 133         h->present                   = 1;
 134     }
 135
 136     return 0;
 137 }
 138
 139 static int decode_registered_user_data_closed_caption(H264SEIA53Caption *h,
 140                                                      GetBitContext *gb, void *logctx,
 141                                                      int size)
 142 {
 143     int flag;
 144     int user_data_type_code;
 145     int cc_count;
 146
 147     if (size < 3)
 148         return AVERROR(EINVAL);
 149
 150     user_data_type_code = get_bits(gb, 8);
 151     if (user_data_type_code == 0x3) {
 152         skip_bits(gb, 1);           // reserved
 153
 154         flag = get_bits(gb, 1);     // process_cc_data_flag
 155         if (flag) {
 156             skip_bits(gb, 1);       // zero bit
 157             cc_count = get_bits(gb, 5);
 158             skip_bits(gb, 8);       // reserved
 159             size -= 2;
 160
 161             if (cc_count && size >= cc_count * 3) {
 162                 const uint64_t new_size = (h->a53_caption_size + cc_count
 163                                            * UINT64_C(3));
 164                 int i, ret;
 165
 166                 if (new_size > INT_MAX)
 167                     return AVERROR(EINVAL);
 168
 169                 /* Allow merging of the cc data from two fields. */
 170                 ret = av_reallocp(&h->a53_caption, new_size);
 171                 if (ret < 0)
 172                     return ret;
 173
 174                 for (i = 0; i < cc_count; i++) {
 175                     h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
 176                     h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
 177                     h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
 178                 }
 179
 180                 skip_bits(gb, 8);   // marker_bits
 181             }
 182         }
 183     } else {
 184         int i;
 185         avpriv_request_sample(logctx, "Subtitles with data type 0x%02x",
 186                               user_data_type_code);
 187         for (i = 0; i < size - 1; i++)
 188             skip_bits(gb, 8);
 189     }
 190
 191     return 0;
 192 }
 193
 194 static int decode_registered_user_data(H264SEIContext *h, GetBitContext *gb,
 195                                        void *logctx, int size)
 196 {
 197     uint32_t country_code;
 198     uint32_t user_identifier;
 199
 200     if (size < 7)
 201         return AVERROR_INVALIDDATA;
 202     size -= 7;
 203
 204     country_code = get_bits(gb, 8); // itu_t_t35_country_code
 205     if (country_code == 0xFF) {
 206         skip_bits(gb, 8);           // itu_t_t35_country_code_extension_byte
 207         size--;
 208     }
 209
 210     /* itu_t_t35_payload_byte follows */
 211     skip_bits(gb, 8);              // terminal provider code
 212     skip_bits(gb, 8);              // terminal provider oriented code
 213     user_identifier = get_bits_long(gb, 32);
 214
 215     switch (user_identifier) {
 216         case MKBETAG('D', 'T', 'G', '1'):       // afd_data
 217             return decode_registered_user_data_afd(&h->afd, gb, size);
 218         case MKBETAG('G', 'A', '9', '4'):       // closed captions
 219             return decode_registered_user_data_closed_caption(&h->a53_caption, gb,
 220                                                               logctx, size);
 221         default:
 222             skip_bits(gb, size * 8);
 223             break;
 224     }
 225
 226     return 0;
 227 }
 228
 229 static int decode_unregistered_user_data(H264SEIUnregistered *h, GetBitContext *gb,
 230                                          void *logctx, int size)
 231 {
 232     uint8_t *user_data;
 233     int e, build, i;
 234
 235     if (size < 16 || size >= INT_MAX - 16)
 236         return AVERROR_INVALIDDATA;
 237
 238     user_data = av_malloc(16 + size + 1);
 239     if (!user_data)
 240         return AVERROR(ENOMEM);
 241
 242     for (i = 0; i < size + 16; i++)
 243         user_data[i] = get_bits(gb, 8);
 244
 245     user_data[i] = 0;
 246     e = sscanf(user_data + 16, "x264 - core %d", &build);
 247     if (e == 1 && build > 0)
 248         h->x264_build = build;
 249
 250     if (strlen(user_data + 16) > 0)
 251         av_log(logctx, AV_LOG_DEBUG, "user data:\"%s\"\n", user_data + 16);
 252
 253     av_free(user_data);
 254     return 0;
 255 }
 256
 257 static int decode_recovery_point(H264SEIRecoveryPoint *h, GetBitContext *gb)
 258 {
 259     h->recovery_frame_cnt = get_ue_golomb(gb);
 260
 261     /* 1b exact_match_flag,
 262      * 1b broken_link_flag,
 263      * 2b changing_slice_group_idc */
 264     skip_bits(gb, 4);
 265
 266     return 0;
 267 }
 268
 269 static int decode_buffering_period(H264SEIBufferingPeriod *h, GetBitContext *gb,
 270                                    const H264ParamSets *ps, void *logctx)
 271 {
 272     unsigned int sps_id;
 273     int sched_sel_idx;
 274     SPS *sps;
 275
 276     sps_id = get_ue_golomb_31(gb);
 277     if (sps_id > 31 || !ps->sps_list[sps_id]) {
 278         av_log(logctx, AV_LOG_ERROR,
 279                "non-existing SPS %d referenced in buffering period\n", sps_id);
 280         return AVERROR_INVALIDDATA;
 281     }
 282     sps = (SPS*)ps->sps_list[sps_id]->data;
 283
 284     // NOTE: This is really so duplicated in the standard... See H.264, D.1.1
 285     if (sps->nal_hrd_parameters_present_flag) {
 286         for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
 287             h->initial_cpb_removal_delay[sched_sel_idx] =
 288                 get_bits(gb, sps->initial_cpb_removal_delay_length);
 289             // initial_cpb_removal_delay_offset
 290             skip_bits(gb, sps->initial_cpb_removal_delay_length);
 291         }
 292     }
 293     if (sps->vcl_hrd_parameters_present_flag) {
 294         for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
 295             h->initial_cpb_removal_delay[sched_sel_idx] =
 296                 get_bits(gb, sps->initial_cpb_removal_delay_length);
 297             // initial_cpb_removal_delay_offset
 298             skip_bits(gb, sps->initial_cpb_removal_delay_length);
 299         }
 300     }
 301
 302     h->present = 1;
 303     return 0;
 304 }
 305
 306 static int decode_frame_packing_arrangement(H264SEIFramePacking *h,
 307                                             GetBitContext *gb)
 308 {
 309     get_ue_golomb(gb);              // frame_packing_arrangement_id
 310     h->present = !get_bits1(gb);
 311
 312     if (h->present) {
 313         h->arrangement_type = get_bits(gb, 7);
 314         h->quincunx_subsampling           = get_bits1(gb);
 315         h->content_interpretation_type    = get_bits(gb, 6);
 316
 317         // spatial_flipping_flag, frame0_flipped_flag, field_views_flag
 318         skip_bits(gb, 3);
 319         h->current_frame_is_frame0_flag = get_bits1(gb);
 320         // frame0_self_contained_flag, frame1_self_contained_flag
 321         skip_bits(gb, 2);
 322
 323         if (!h->quincunx_subsampling && h->arrangement_type != 5)
 324             skip_bits(gb, 16);      // frame[01]_grid_position_[xy]
 325         skip_bits(gb, 8);           // frame_packing_arrangement_reserved_byte
 326         get_ue_golomb(gb);          // frame_packing_arrangement_repetition_period
 327     }
 328     skip_bits1(gb);                 // frame_packing_arrangement_extension_flag
 329
 330     return 0;
 331 }
 332
 333 static int decode_display_orientation(H264SEIDisplayOrientation *h,
 334                                       GetBitContext *gb)
 335 {
 336     h->present = !get_bits1(gb);
 337
 338     if (h->present) {
 339         h->hflip = get_bits1(gb);     // hor_flip
 340         h->vflip = get_bits1(gb);     // ver_flip
 341
 342         h->anticlockwise_rotation = get_bits(gb, 16);
 343         get_ue_golomb(gb);  // display_orientation_repetition_period
 344         skip_bits1(gb);     // display_orientation_extension_flag
 345     }
 346
 347     return 0;
 348 }
 349
 350 static int decode_alternative_transfer(H264SEIAlternativeTransfer *h,
 351                                        GetBitContext *gb)
 352 {
 353     h->present = 1;
 354     h->preferred_transfer_characteristics = get_bits(gb, 8);
 355     return 0;
 356 }
 357
 358 int ff_h264_sei_decode(H264SEIContext *h, GetBitContext *gb,
 359                        const H264ParamSets *ps, void *logctx)
 360 {
 361     while (get_bits_left(gb) > 16) {
 362         int size = 0;
 363         int type = 0;
 364         int ret  = 0;
 365         int last = 0;
 366
 367         while (get_bits_left(gb) >= 8 &&
 368                (last = get_bits(gb, 8)) == 255) {
 369             type += 255;
 370         }
 371         type += last;
 372
 373         last = 0;
 374         while (get_bits_left(gb) >= 8 &&
 375                (last = get_bits(gb, 8)) == 255) {
 376             size += 255;
 377         }
 378         size += last;
 379
 380         if (size > get_bits_left(gb) / 8) {
 381             av_log(logctx, AV_LOG_ERROR, "SEI type %d truncated at %d\n",
 382                    type, get_bits_left(gb));
 383             return AVERROR_INVALIDDATA;
 384         }
 385
 386         switch (type) {
 387         case H264_SEI_TYPE_PIC_TIMING: // Picture timing SEI
 388             ret = decode_picture_timing(&h->picture_timing, gb, ps->sps, logctx);
 389             break;
 390         case H264_SEI_TYPE_USER_DATA_REGISTERED:
 391             ret = decode_registered_user_data(h, gb, logctx, size);
 392             break;
 393         case H264_SEI_TYPE_USER_DATA_UNREGISTERED:
 394             ret = decode_unregistered_user_data(&h->unregistered, gb, logctx, size);
 395             break;
 396         case H264_SEI_TYPE_RECOVERY_POINT:
 397             ret = decode_recovery_point(&h->recovery_point, gb);
 398             break;
 399         case H264_SEI_TYPE_BUFFERING_PERIOD:
 400             ret = decode_buffering_period(&h->buffering_period, gb, ps, logctx);
 401             break;
 402         case H264_SEI_TYPE_FRAME_PACKING:
 403             ret = decode_frame_packing_arrangement(&h->frame_packing, gb);
 404             break;
 405         case H264_SEI_TYPE_DISPLAY_ORIENTATION:
 406             ret = decode_display_orientation(&h->display_orientation, gb);
 407             break;
 408         case H264_SEI_TYPE_ALTERNATIVE_TRANSFER:
 409             ret = decode_alternative_transfer(&h->alternative_transfer, gb);
 410             break;
 411         default:
 412             av_log(logctx, AV_LOG_DEBUG, "unknown SEI type %d\n", type);
 413             skip_bits(gb, 8 * size);
 414         }
 415         if (ret < 0)
 416             return ret;
 417
 418         // FIXME check bits here
 419         align_get_bits(gb);
 420     }
 421
 422     return 0;
 423 }