libavformat/oggenc.c

   1 /*
   2  * Ogg muxer
   3  * Copyright (c) 2007 Baptiste Coudurier <baptiste dot coudurier at free dot fr>
   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 "libavutil/crc.h"
  23 #include "libavcodec/xiph.h"
  24 #include "libavcodec/bytestream.h"
  25 #include "libavcodec/flac.h"
  26 #include "avformat.h"
  27 #include "internal.h"
  28
  29 typedef struct {
  30     int64_t duration;
  31     unsigned page_counter;
  32     uint8_t *header[3];
  33     int header_len[3];
  34     /** for theora granule */
  35     int kfgshift;
  36     int64_t last_kf_pts;
  37     int vrev;
  38     int eos;
  39 } OGGStreamContext;
  40
  41 static void ogg_update_checksum(AVFormatContext *s, int64_t crc_offset)
  42 {
  43     int64_t pos = url_ftell(s->pb);
  44     uint32_t checksum = get_checksum(s->pb);
  45     url_fseek(s->pb, crc_offset, SEEK_SET);
  46     put_be32(s->pb, checksum);
  47     url_fseek(s->pb, pos, SEEK_SET);
  48 }
  49
  50 static int ogg_write_page(AVFormatContext *s, const uint8_t *data, int size,
  51                           int64_t granule, int stream_index, int flags)
  52 {
  53     OGGStreamContext *oggstream = s->streams[stream_index]->priv_data;
  54     int64_t crc_offset;
  55     int page_segments, i;
  56
  57     if (size >= 255*255) {
  58         granule = -1;
  59         size = 255*255;
  60     } else if (oggstream->eos)
  61         flags |= 4;
  62
  63     page_segments = FFMIN((size/255)+!!size, 255);
  64
  65     init_checksum(s->pb, ff_crc04C11DB7_update, 0);
  66     put_tag(s->pb, "OggS");
  67     put_byte(s->pb, 0);
  68     put_byte(s->pb, flags);
  69     put_le64(s->pb, granule);
  70     put_le32(s->pb, stream_index);
  71     put_le32(s->pb, oggstream->page_counter++);
  72     crc_offset = url_ftell(s->pb);
  73     put_le32(s->pb, 0); // crc
  74     put_byte(s->pb, page_segments);
  75     for (i = 0; i < page_segments-1; i++)
  76         put_byte(s->pb, 255);
  77     if (size) {
  78         put_byte(s->pb, size - (page_segments-1)*255);
  79         put_buffer(s->pb, data, size);
  80     }
  81     ogg_update_checksum(s, crc_offset);
  82     put_flush_packet(s->pb);
  83     return size;
  84 }
  85
  86 static uint8_t *ogg_write_vorbiscomment(int offset, int bitexact,
  87                                         int *header_len)
  88 {
  89     const char *vendor = bitexact ? "ffmpeg" : LIBAVFORMAT_IDENT;
  90     int size;
  91     uint8_t *p, *p0;
  92
  93     size = offset + 4 + strlen(vendor) + 4;
  94     p = av_mallocz(size);
  95     if (!p)
  96         return NULL;
  97     p0 = p;
  98
  99     p += offset;
 100     bytestream_put_le32(&p, strlen(vendor));
 101     bytestream_put_buffer(&p, vendor, strlen(vendor));
 102     bytestream_put_le32(&p, 0); // user comment list length
 103
 104     *header_len = size;
 105     return p0;
 106 }
 107
 108 static int ogg_build_flac_headers(AVCodecContext *avctx,
 109                                   OGGStreamContext *oggstream, int bitexact)
 110 {
 111     enum FLACExtradataFormat format;
 112     uint8_t *streaminfo;
 113     uint8_t *p;
 114
 115     if (!ff_flac_is_extradata_valid(avctx, &format, &streaminfo))
 116         return -1;
 117
 118     // first packet: STREAMINFO
 119     oggstream->header_len[0] = 51;
 120     oggstream->header[0] = av_mallocz(51); // per ogg flac specs
 121     p = oggstream->header[0];
 122     if (!p)
 123         return AVERROR_NOMEM;
 124     bytestream_put_byte(&p, 0x7F);
 125     bytestream_put_buffer(&p, "FLAC", 4);
 126     bytestream_put_byte(&p, 1); // major version
 127     bytestream_put_byte(&p, 0); // minor version
 128     bytestream_put_be16(&p, 1); // headers packets without this one
 129     bytestream_put_buffer(&p, "fLaC", 4);
 130     bytestream_put_byte(&p, 0x00); // streaminfo
 131     bytestream_put_be24(&p, 34);
 132     bytestream_put_buffer(&p, streaminfo, FLAC_STREAMINFO_SIZE);
 133
 134     // second packet: VorbisComment
 135     p = ogg_write_vorbiscomment(4, bitexact, &oggstream->header_len[1]);
 136     if (!p)
 137         return AVERROR_NOMEM;
 138     oggstream->header[1] = p;
 139     bytestream_put_byte(&p, 0x84); // last metadata block and vorbis comment
 140     bytestream_put_be24(&p, oggstream->header_len[1] - 4);
 141
 142     return 0;
 143 }
 144
 145 #define SPEEX_HEADER_SIZE 80
 146
 147 static int ogg_build_speex_headers(AVCodecContext *avctx,
 148                                    OGGStreamContext *oggstream, int bitexact)
 149 {
 150     uint8_t *p;
 151
 152     if (avctx->extradata_size < SPEEX_HEADER_SIZE)
 153         return -1;
 154
 155     // first packet: Speex header
 156     p = av_mallocz(SPEEX_HEADER_SIZE);
 157     if (!p)
 158         return AVERROR_NOMEM;
 159     oggstream->header[0] = p;
 160     oggstream->header_len[0] = SPEEX_HEADER_SIZE;
 161     bytestream_put_buffer(&p, avctx->extradata, SPEEX_HEADER_SIZE);
 162     AV_WL32(&oggstream->header[0][68], 0);  // set extra_headers to 0
 163
 164     // second packet: VorbisComment
 165     p = ogg_write_vorbiscomment(0, bitexact, &oggstream->header_len[1]);
 166     if (!p)
 167         return AVERROR_NOMEM;
 168     oggstream->header[1] = p;
 169
 170     return 0;
 171 }
 172
 173 static int ogg_write_header(AVFormatContext *s)
 174 {
 175     OGGStreamContext *oggstream;
 176     int i, j;
 177     for (i = 0; i < s->nb_streams; i++) {
 178         AVStream *st = s->streams[i];
 179         if (st->codec->codec_type == CODEC_TYPE_AUDIO)
 180             av_set_pts_info(st, 64, 1, st->codec->sample_rate);
 181         else if (st->codec->codec_type == CODEC_TYPE_VIDEO)
 182             av_set_pts_info(st, 64, st->codec->time_base.num, st->codec->time_base.den);
 183         if (st->codec->codec_id != CODEC_ID_VORBIS &&
 184             st->codec->codec_id != CODEC_ID_THEORA &&
 185             st->codec->codec_id != CODEC_ID_SPEEX  &&
 186             st->codec->codec_id != CODEC_ID_FLAC) {
 187             av_log(s, AV_LOG_ERROR, "Unsupported codec id in stream %d\n", i);
 188             return -1;
 189         }
 190
 191         if (!st->codec->extradata || !st->codec->extradata_size) {
 192             av_log(s, AV_LOG_ERROR, "No extradata present\n");
 193             return -1;
 194         }
 195         oggstream = av_mallocz(sizeof(*oggstream));
 196         st->priv_data = oggstream;
 197         if (st->codec->codec_id == CODEC_ID_FLAC) {
 198             int err = ogg_build_flac_headers(st->codec, oggstream,
 199                                              st->codec->flags & CODEC_FLAG_BITEXACT);
 200             if (err) {
 201                 av_log(s, AV_LOG_ERROR, "Error writing FLAC headers\n");
 202                 av_freep(&st->priv_data);
 203                 return err;
 204             }
 205         } else if (st->codec->codec_id == CODEC_ID_SPEEX) {
 206             int err = ogg_build_speex_headers(st->codec, oggstream,
 207                                               st->codec->flags & CODEC_FLAG_BITEXACT);
 208             if (err) {
 209                 av_log(s, AV_LOG_ERROR, "Error writing Speex headers\n");
 210                 av_freep(&st->priv_data);
 211                 return err;
 212             }
 213         } else {
 214             if (ff_split_xiph_headers(st->codec->extradata, st->codec->extradata_size,
 215                                       st->codec->codec_id == CODEC_ID_VORBIS ? 30 : 42,
 216                                       oggstream->header, oggstream->header_len) < 0) {
 217                 av_log(s, AV_LOG_ERROR, "Extradata corrupted\n");
 218                 av_freep(&st->priv_data);
 219                 return -1;
 220             }
 221             if (st->codec->codec_id == CODEC_ID_THEORA) {
 222                 /** KFGSHIFT is the width of the less significant section of the granule position
 223                     The less significant section is the frame count since the last keyframe */
 224                 oggstream->kfgshift = ((oggstream->header[0][40]&3)<<3)|(oggstream->header[0][41]>>5);
 225                 oggstream->vrev = oggstream->header[0][9];
 226                 av_log(s, AV_LOG_DEBUG, "theora kfgshift %d, vrev %d\n",
 227                        oggstream->kfgshift, oggstream->vrev);
 228             }
 229         }
 230     }
 231     for (i = 0; i < 3; i++) {
 232         for (j = 0; j < s->nb_streams; j++) {
 233             AVStream *st = s->streams[j];
 234             OGGStreamContext *oggstream = st->priv_data;
 235             if (oggstream && oggstream->header_len[i]) {
 236                 ogg_write_page(s, oggstream->header[i], oggstream->header_len[i],
 237                                0, st->index, i ? 0 : 2); // bos
 238             }
 239         }
 240     }
 241     return 0;
 242 }
 243
 244 static int ogg_write_packet(AVFormatContext *s, AVPacket *pkt)
 245 {
 246     AVStream *st = s->streams[pkt->stream_index];
 247     OGGStreamContext *oggstream = st->priv_data;
 248     uint8_t *ptr = pkt->data;
 249     int ret, size = pkt->size;
 250     int64_t granule;
 251
 252     if (st->codec->codec_id == CODEC_ID_THEORA) {
 253         int64_t pts = oggstream->vrev < 1 ? pkt->pts : pkt->pts + pkt->duration;
 254         int pframe_count;
 255         if (pkt->flags & PKT_FLAG_KEY)
 256             oggstream->last_kf_pts = pts;
 257         pframe_count = pts - oggstream->last_kf_pts;
 258         // prevent frame count from overflow if key frame flag is not set
 259         if (pframe_count >= (1<<oggstream->kfgshift)) {
 260             oggstream->last_kf_pts += pframe_count;
 261             pframe_count = 0;
 262         }
 263         granule = (oggstream->last_kf_pts<<oggstream->kfgshift) | pframe_count;
 264     } else
 265         granule = pkt->pts + pkt->duration;
 266     oggstream->duration = granule;
 267     do {
 268         ret = ogg_write_page(s, ptr, size, granule, pkt->stream_index, ptr != pkt->data);
 269         ptr  += ret; size -= ret;
 270     } while (size > 0 || ret == 255*255); // need to output a last nil page
 271
 272     return 0;
 273 }
 274
 275 static int ogg_compare_granule(AVFormatContext *s, AVPacket *next, AVPacket *pkt)
 276 {
 277     AVStream *st2 = s->streams[next->stream_index];
 278     AVStream *st  = s->streams[pkt ->stream_index];
 279
 280     int64_t next_granule = av_rescale_q(next->pts + next->duration,
 281                                         st2->time_base, AV_TIME_BASE_Q);
 282     int64_t cur_granule  = av_rescale_q(pkt ->pts + pkt ->duration,
 283                                         st ->time_base, AV_TIME_BASE_Q);
 284     return next_granule > cur_granule;
 285 }
 286
 287 static int ogg_interleave_per_granule(AVFormatContext *s, AVPacket *out, AVPacket *pkt, int flush)
 288 {
 289     AVPacketList *pktl;
 290     int stream_count = 0;
 291     int streams[MAX_STREAMS] = {0};
 292     int interleaved = 0;
 293
 294     if (pkt) {
 295         ff_interleave_add_packet(s, pkt, ogg_compare_granule);
 296     }
 297
 298     pktl = s->packet_buffer;
 299     while (pktl) {
 300         if (streams[pktl->pkt.stream_index] == 0)
 301             stream_count++;
 302         streams[pktl->pkt.stream_index]++;
 303         // need to buffer at least one packet to set eos flag
 304         if (streams[pktl->pkt.stream_index] == 2)
 305             interleaved++;
 306         pktl = pktl->next;
 307     }
 308
 309     if ((s->nb_streams == stream_count && interleaved == stream_count) ||
 310         (flush && stream_count)) {
 311         pktl= s->packet_buffer;
 312         *out= pktl->pkt;
 313         s->packet_buffer = pktl->next;
 314         if (flush && streams[out->stream_index] == 1) {
 315             OGGStreamContext *ogg = s->streams[out->stream_index]->priv_data;
 316             ogg->eos = 1;
 317         }
 318         if(!s->packet_buffer)
 319             s->packet_buffer_end= NULL;
 320
 321         if(s->streams[out->stream_index]->last_in_packet_buffer == pktl)
 322             s->streams[out->stream_index]->last_in_packet_buffer= NULL;
 323
 324         av_freep(&pktl);
 325         return 1;
 326     } else {
 327         av_init_packet(out);
 328         return 0;
 329     }
 330 }
 331
 332 static int ogg_write_trailer(AVFormatContext *s)
 333 {
 334     int i;
 335     for (i = 0; i < s->nb_streams; i++) {
 336         AVStream *st = s->streams[i];
 337         OGGStreamContext *oggstream = st->priv_data;
 338         if (st->codec->codec_id == CODEC_ID_FLAC ||
 339             st->codec->codec_id == CODEC_ID_SPEEX) {
 340             av_free(oggstream->header[0]);
 341             av_free(oggstream->header[1]);
 342         }
 343         av_freep(&st->priv_data);
 344     }
 345     return 0;
 346 }
 347
 348 AVOutputFormat ogg_muxer = {
 349     "ogg",
 350     NULL_IF_CONFIG_SMALL("Ogg"),
 351     "application/ogg",
 352     "ogg,ogv,spx",
 353     0,
 354     CODEC_ID_FLAC,
 355     CODEC_ID_THEORA,
 356     ogg_write_header,
 357     ogg_write_packet,
 358     ogg_write_trailer,
 359     .interleave_packet = ogg_interleave_per_granule,
 360 };