libavformat/av1dec.c

   1 /*
   2  * AV1 Annex B demuxer
   3  * Copyright (c) 2019 James Almer <jamrial@gmail.com>
   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 #include "config_components.h"
  23
  24 #include "libavutil/common.h"
  25 #include "libavutil/opt.h"
  26 #include "libavcodec/av1_parse.h"
  27 #include "libavcodec/bsf.h"
  28 #include "avformat.h"
  29 #include "avio_internal.h"
  30 #include "demux.h"
  31 #include "internal.h"
  32
  33 typedef struct AV1DemuxContext {
  34     const AVClass *class;
  35     AVBSFContext *bsf;
  36     AVRational framerate;
  37     uint32_t temporal_unit_size;
  38     uint32_t frame_unit_size;
  39 } AV1DemuxContext;
  40
  41 //return < 0 if we need more data
  42 static int get_score(int type, int *seq)
  43 {
  44     switch (type) {
  45     case AV1_OBU_SEQUENCE_HEADER:
  46         *seq = 1;
  47         return -1;
  48     case AV1_OBU_FRAME:
  49     case AV1_OBU_FRAME_HEADER:
  50         return *seq ? AVPROBE_SCORE_EXTENSION + 1 : 0;
  51     case AV1_OBU_METADATA:
  52     case AV1_OBU_PADDING:
  53         return -1;
  54     default:
  55         break;
  56     }
  57     return 0;
  58 }
  59
  60 static int av1_read_header(AVFormatContext *s)
  61 {
  62     AV1DemuxContext *const c = s->priv_data;
  63     const AVBitStreamFilter *filter = av_bsf_get_by_name("av1_frame_merge");
  64     AVStream *st;
  65     FFStream *sti;
  66     int ret;
  67
  68     if (!filter) {
  69         av_log(s, AV_LOG_ERROR, "av1_frame_merge bitstream filter "
  70                "not found. This is a bug, please report it.\n");
  71         return AVERROR_BUG;
  72     }
  73
  74     st = avformat_new_stream(s, NULL);
  75     if (!st)
  76         return AVERROR(ENOMEM);
  77     sti = ffstream(st);
  78
  79     st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
  80     st->codecpar->codec_id = AV_CODEC_ID_AV1;
  81     sti->need_parsing = AVSTREAM_PARSE_HEADERS;
  82
  83     st->avg_frame_rate = c->framerate;
  84     // taken from rawvideo demuxers
  85     avpriv_set_pts_info(st, 64, 1, 1200000);
  86
  87     ret = av_bsf_alloc(filter, &c->bsf);
  88     if (ret < 0)
  89         return ret;
  90
  91     ret = avcodec_parameters_copy(c->bsf->par_in, st->codecpar);
  92     if (ret < 0)
  93         return ret;
  94
  95     ret = av_bsf_init(c->bsf);
  96     if (ret < 0)
  97         return ret;
  98
  99     return 0;
 100 }
 101
 102 static int av1_read_close(AVFormatContext *s)
 103 {
 104     AV1DemuxContext *const c = s->priv_data;
 105
 106     av_bsf_free(&c->bsf);
 107     return 0;
 108 }
 109
 110 #define DEC AV_OPT_FLAG_DECODING_PARAM
 111 #define OFFSET(x) offsetof(AV1DemuxContext, x)
 112 static const AVOption av1_options[] = {
 113     { "framerate", "", OFFSET(framerate), AV_OPT_TYPE_VIDEO_RATE, {.str = "25"}, 0, INT_MAX, DEC},
 114     { NULL },
 115 };
 116 #undef OFFSET
 117
 118 static const AVClass av1_demuxer_class = {
 119     .class_name = "AV1 Annex B/low overhead OBU demuxer",
 120     .item_name  = av_default_item_name,
 121     .option     = av1_options,
 122     .version    = LIBAVUTIL_VERSION_INT,
 123 };
 124
 125 #if CONFIG_AV1_DEMUXER
 126
 127 static int leb(AVIOContext *pb, uint32_t *len, int eof) {
 128     int more, i = 0;
 129     *len = 0;
 130     do {
 131         unsigned bits;
 132         int byte = avio_r8(pb);
 133         if (pb->error)
 134             return pb->error;
 135         if (pb->eof_reached)
 136             return (eof && !i) ? AVERROR_EOF : AVERROR_INVALIDDATA;
 137         more = byte & 0x80;
 138         bits = byte & 0x7f;
 139         if (i <= 3 || (i == 4 && bits < (1 << 4)))
 140             *len |= bits << (i * 7);
 141         else if (bits)
 142             return AVERROR_INVALIDDATA;
 143         if (++i == 8 && more)
 144             return AVERROR_INVALIDDATA;
 145     } while (more);
 146     return i;
 147 }
 148
 149 static int read_obu(const uint8_t *buf, int size, int64_t *obu_size, int *type)
 150 {
 151     int start_pos, temporal_id, spatial_id;
 152     int len;
 153
 154     len = parse_obu_header(buf, size, obu_size, &start_pos,
 155                            type, &temporal_id, &spatial_id);
 156     if (len < 0)
 157         return len;
 158
 159     return 0;
 160 }
 161
 162 static int annexb_probe(const AVProbeData *p)
 163 {
 164     FFIOContext ctx;
 165     AVIOContext *const pb = &ctx.pub;
 166     int64_t obu_size;
 167     uint32_t temporal_unit_size, frame_unit_size, obu_unit_size;
 168     int seq = 0;
 169     int ret, type, cnt = 0;
 170
 171     ffio_init_read_context(&ctx, p->buf, p->buf_size);
 172
 173     ret = leb(pb, &temporal_unit_size, 1);
 174     if (ret < 0)
 175         return 0;
 176     cnt += ret;
 177     ret = leb(pb, &frame_unit_size, 0);
 178     if (ret < 0 || ((int64_t)frame_unit_size + ret) > temporal_unit_size)
 179         return 0;
 180     cnt += ret;
 181     ret = leb(pb, &obu_unit_size, 0);
 182     if (ret < 0 || ((int64_t)obu_unit_size + ret) >= frame_unit_size)
 183         return 0;
 184     cnt += ret;
 185
 186     frame_unit_size -= obu_unit_size + ret;
 187
 188     avio_skip(pb, obu_unit_size);
 189     if (pb->eof_reached || pb->error)
 190         return 0;
 191
 192     // Check that the first OBU is a Temporal Delimiter.
 193     ret = read_obu(p->buf + cnt, FFMIN(p->buf_size - cnt, obu_unit_size), &obu_size, &type);
 194     if (ret < 0 || type != AV1_OBU_TEMPORAL_DELIMITER || obu_size > 0)
 195         return 0;
 196     cnt += obu_unit_size;
 197
 198     do {
 199         ret = leb(pb, &obu_unit_size, 0);
 200         if (ret < 0 || ((int64_t)obu_unit_size + ret) > frame_unit_size)
 201             return 0;
 202         cnt += ret;
 203
 204         avio_skip(pb, obu_unit_size);
 205         if (pb->eof_reached || pb->error)
 206             return 0;
 207
 208         ret = read_obu(p->buf + cnt, FFMIN(p->buf_size - cnt, obu_unit_size), &obu_size, &type);
 209         if (ret < 0)
 210             return 0;
 211         cnt += obu_unit_size;
 212
 213         ret = get_score(type, &seq);
 214         if (ret >= 0)
 215             return ret;
 216
 217         frame_unit_size -= obu_unit_size + ret;
 218     } while (frame_unit_size);
 219
 220     return 0;
 221 }
 222
 223 static int annexb_read_packet(AVFormatContext *s, AVPacket *pkt)
 224 {
 225     AV1DemuxContext *const c = s->priv_data;
 226     uint32_t obu_unit_size;
 227     int ret, len;
 228
 229 retry:
 230     if (avio_feof(s->pb)) {
 231         if (c->temporal_unit_size || c->frame_unit_size)
 232             return AVERROR_INVALIDDATA;
 233         goto end;
 234     }
 235
 236     if (!c->temporal_unit_size) {
 237         len = leb(s->pb, &c->temporal_unit_size, 1);
 238         if (len == AVERROR_EOF) goto end;
 239         else if (len < 0) return len;
 240     }
 241
 242     if (!c->frame_unit_size) {
 243         len = leb(s->pb, &c->frame_unit_size, 0);
 244         if (len < 0)
 245             return len;
 246         if (((int64_t)c->frame_unit_size + len) > c->temporal_unit_size)
 247             return AVERROR_INVALIDDATA;
 248         c->temporal_unit_size -= len;
 249     }
 250
 251     len = leb(s->pb, &obu_unit_size, 0);
 252     if (len < 0)
 253         return len;
 254     if (((int64_t)obu_unit_size + len) > c->frame_unit_size)
 255         return AVERROR_INVALIDDATA;
 256
 257     ret = av_get_packet(s->pb, pkt, obu_unit_size);
 258     if (ret < 0)
 259         return ret;
 260     if (ret != obu_unit_size)
 261         return AVERROR_INVALIDDATA;
 262
 263     c->temporal_unit_size -= obu_unit_size + len;
 264     c->frame_unit_size -= obu_unit_size + len;
 265
 266 end:
 267     ret = av_bsf_send_packet(c->bsf, pkt);
 268     if (ret < 0) {
 269         av_log(s, AV_LOG_ERROR, "Failed to send packet to "
 270                                 "av1_frame_merge filter\n");
 271         return ret;
 272     }
 273
 274     ret = av_bsf_receive_packet(c->bsf, pkt);
 275     if (ret < 0 && ret != AVERROR(EAGAIN) && ret != AVERROR_EOF)
 276         av_log(s, AV_LOG_ERROR, "av1_frame_merge filter failed to "
 277                                 "send output packet\n");
 278
 279     if (ret == AVERROR(EAGAIN))
 280         goto retry;
 281
 282     return ret;
 283 }
 284
 285 const FFInputFormat ff_av1_demuxer = {
 286     .p.name         = "av1",
 287     .p.long_name    = NULL_IF_CONFIG_SMALL("AV1 Annex B"),
 288     .p.extensions   = "obu",
 289     .p.flags        = AVFMT_GENERIC_INDEX | AVFMT_NOTIMESTAMPS,
 290     .p.priv_class   = &av1_demuxer_class,
 291     .priv_data_size = sizeof(AV1DemuxContext),
 292     .flags_internal = FF_INFMT_FLAG_INIT_CLEANUP,
 293     .read_probe     = annexb_probe,
 294     .read_header    = av1_read_header,
 295     .read_packet    = annexb_read_packet,
 296     .read_close     = av1_read_close,
 297 };
 298 #endif
 299
 300 #if CONFIG_OBU_DEMUXER
 301 //For low overhead obu, we can't foresee the obu size before we parsed the header.
 302 //So, we can't use parse_obu_header here, since it will check size <= buf_size
 303 //see c27c7b49dc for more details
 304 static int read_obu_with_size(const uint8_t *buf, int buf_size, int64_t *obu_size, int *type)
 305 {
 306     GetBitContext gb;
 307     int ret, extension_flag, start_pos;
 308     int64_t size;
 309
 310     ret = init_get_bits8(&gb, buf, FFMIN(buf_size, MAX_OBU_HEADER_SIZE));
 311     if (ret < 0)
 312         return ret;
 313
 314     if (get_bits1(&gb) != 0) // obu_forbidden_bit
 315         return AVERROR_INVALIDDATA;
 316
 317     *type      = get_bits(&gb, 4);
 318     extension_flag = get_bits1(&gb);
 319     if (!get_bits1(&gb))    // has_size_flag
 320         return AVERROR_INVALIDDATA;
 321     skip_bits1(&gb);        // obu_reserved_1bit
 322
 323     if (extension_flag) {
 324         get_bits(&gb, 3);   // temporal_id
 325         get_bits(&gb, 2);   // spatial_id
 326         skip_bits(&gb, 3);  // extension_header_reserved_3bits
 327     }
 328
 329     *obu_size  = get_leb128(&gb);
 330     if (*obu_size > INT_MAX)
 331         return AVERROR_INVALIDDATA;
 332
 333     if (get_bits_left(&gb) < 0)
 334         return AVERROR_INVALIDDATA;
 335
 336     start_pos = get_bits_count(&gb) / 8;
 337
 338     size = *obu_size + start_pos;
 339     if (size > INT_MAX)
 340         return AVERROR_INVALIDDATA;
 341     return size;
 342 }
 343
 344 static int obu_probe(const AVProbeData *p)
 345 {
 346     int64_t obu_size;
 347     int seq = 0;
 348     int ret, type, cnt;
 349
 350     // Check that the first OBU is a Temporal Delimiter.
 351     cnt = read_obu_with_size(p->buf, p->buf_size, &obu_size, &type);
 352     if (cnt < 0 || type != AV1_OBU_TEMPORAL_DELIMITER || obu_size != 0)
 353         return 0;
 354
 355     while (1) {
 356         ret = read_obu_with_size(p->buf + cnt, p->buf_size - cnt, &obu_size, &type);
 357         if (ret < 0 || obu_size <= 0)
 358             return 0;
 359         cnt += FFMIN(ret, p->buf_size - cnt);
 360
 361         ret = get_score(type, &seq);
 362         if (ret >= 0)
 363             return ret;
 364     }
 365     return 0;
 366 }
 367
 368 static int obu_get_packet(AVFormatContext *s, AVPacket *pkt)
 369 {
 370     AV1DemuxContext *const c = s->priv_data;
 371     uint8_t header[MAX_OBU_HEADER_SIZE + AV_INPUT_BUFFER_PADDING_SIZE];
 372     int64_t obu_size;
 373     int size;
 374     int ret, len, type;
 375
 376     if ((ret = ffio_ensure_seekback(s->pb, MAX_OBU_HEADER_SIZE)) < 0)
 377         return ret;
 378     size = avio_read(s->pb, header, MAX_OBU_HEADER_SIZE);
 379     if (size < 0)
 380         return size;
 381
 382     memset(header + size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
 383     len = read_obu_with_size(header, size, &obu_size, &type);
 384     if (len < 0) {
 385         av_log(c, AV_LOG_ERROR, "Failed to read obu\n");
 386         return len;
 387     }
 388     avio_seek(s->pb, -size, SEEK_CUR);
 389
 390     ret = av_get_packet(s->pb, pkt, len);
 391     if (ret != len) {
 392         av_log(c, AV_LOG_ERROR, "Failed to get packet for obu\n");
 393         return ret < 0 ? ret : AVERROR_INVALIDDATA;
 394     }
 395     return 0;
 396 }
 397
 398 static int obu_read_packet(AVFormatContext *s, AVPacket *pkt)
 399 {
 400     AV1DemuxContext *const c = s->priv_data;
 401     int ret;
 402
 403     if (s->io_repositioned) {
 404         av_bsf_flush(c->bsf);
 405         s->io_repositioned = 0;
 406     }
 407     while (1) {
 408         ret = obu_get_packet(s, pkt);
 409         /* In case of AVERROR_EOF we need to flush the BSF. Conveniently
 410          * obu_get_packet() returns a blank pkt in this case which
 411          * can be used to signal that the BSF should be flushed. */
 412         if (ret < 0 && ret != AVERROR_EOF)
 413             return ret;
 414         ret = av_bsf_send_packet(c->bsf, pkt);
 415         if (ret < 0) {
 416             av_log(s, AV_LOG_ERROR, "Failed to send packet to "
 417                                     "av1_frame_merge filter\n");
 418             return ret;
 419         }
 420         ret = av_bsf_receive_packet(c->bsf, pkt);
 421         if (ret < 0 && ret != AVERROR(EAGAIN) && ret != AVERROR_EOF)
 422             av_log(s, AV_LOG_ERROR, "av1_frame_merge filter failed to "
 423                                     "send output packet\n");
 424         if (ret != AVERROR(EAGAIN))
 425             break;
 426     }
 427
 428     return ret;
 429 }
 430
 431 const FFInputFormat ff_obu_demuxer = {
 432     .p.name         = "obu",
 433     .p.long_name    = NULL_IF_CONFIG_SMALL("AV1 low overhead OBU"),
 434     .p.extensions   = "obu",
 435     .p.flags        = AVFMT_GENERIC_INDEX | AVFMT_NO_BYTE_SEEK | AVFMT_NOTIMESTAMPS,
 436     .p.priv_class   = &av1_demuxer_class,
 437     .priv_data_size = sizeof(AV1DemuxContext),
 438     .flags_internal = FF_INFMT_FLAG_INIT_CLEANUP,
 439     .read_probe     = obu_probe,
 440     .read_header    = av1_read_header,
 441     .read_packet    = obu_read_packet,
 442     .read_close     = av1_read_close,
 443 };
 444 #endif