libavformat/flvdec.c

   1 /*
   2  * FLV demuxer
   3  * Copyright (c) 2003 The FFmpeg Project
   4  *
   5  * This demuxer will generate a 1 byte extradata for VP6F content.
   6  * It is composed of:
   7  *  - upper 4bits: difference between encoded width and visible width
   8  *  - lower 4bits: difference between encoded height and visible height
   9  *
  10  * This file is part of FFmpeg.
  11  *
  12  * FFmpeg is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU Lesser General Public
  14  * License as published by the Free Software Foundation; either
  15  * version 2.1 of the License, or (at your option) any later version.
  16  *
  17  * FFmpeg is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20  * Lesser General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU Lesser General Public
  23  * License along with FFmpeg; if not, write to the Free Software
  24  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  25  */
  26
  27 #include "libavcodec/mpeg4audio.h"
  28 #include "avformat.h"
  29 #include "flv.h"
  30
  31 typedef struct {
  32     int wrong_dts; ///< wrong dts due to negative cts
  33 } FLVContext;
  34
  35 static int flv_probe(AVProbeData *p)
  36 {
  37     const uint8_t *d;
  38
  39     d = p->buf;
  40     if (d[0] == 'F' && d[1] == 'L' && d[2] == 'V' && d[3] < 5 && d[5]==0) {
  41         return AVPROBE_SCORE_MAX;
  42     }
  43     return 0;
  44 }
  45
  46 static void flv_set_audio_codec(AVFormatContext *s, AVStream *astream, int flv_codecid) {
  47     AVCodecContext *acodec = astream->codec;
  48     switch(flv_codecid) {
  49         //no distinction between S16 and S8 PCM codec flags
  50         case FLV_CODECID_PCM:
  51             acodec->codec_id = acodec->bits_per_coded_sample == 8 ? CODEC_ID_PCM_S8 :
  52 #ifdef WORDS_BIGENDIAN
  53                                 CODEC_ID_PCM_S16BE;
  54 #else
  55                                 CODEC_ID_PCM_S16LE;
  56 #endif
  57             break;
  58         case FLV_CODECID_PCM_LE:
  59             acodec->codec_id = acodec->bits_per_coded_sample == 8 ? CODEC_ID_PCM_S8 : CODEC_ID_PCM_S16LE; break;
  60         case FLV_CODECID_AAC  : acodec->codec_id = CODEC_ID_AAC;                                    break;
  61         case FLV_CODECID_ADPCM: acodec->codec_id = CODEC_ID_ADPCM_SWF;                              break;
  62         case FLV_CODECID_SPEEX:
  63             acodec->codec_id = CODEC_ID_SPEEX;
  64             acodec->sample_rate = 16000;
  65             break;
  66         case FLV_CODECID_MP3  : acodec->codec_id = CODEC_ID_MP3      ; astream->need_parsing = AVSTREAM_PARSE_FULL; break;
  67         case FLV_CODECID_NELLYMOSER_8KHZ_MONO:
  68             acodec->sample_rate = 8000; //in case metadata does not otherwise declare samplerate
  69         case FLV_CODECID_NELLYMOSER:
  70             acodec->codec_id = CODEC_ID_NELLYMOSER;
  71             break;
  72         default:
  73             av_log(s, AV_LOG_INFO, "Unsupported audio codec (%x)\n", flv_codecid >> FLV_AUDIO_CODECID_OFFSET);
  74             acodec->codec_tag = flv_codecid >> FLV_AUDIO_CODECID_OFFSET;
  75     }
  76 }
  77
  78 static int flv_set_video_codec(AVFormatContext *s, AVStream *vstream, int flv_codecid) {
  79     AVCodecContext *vcodec = vstream->codec;
  80     switch(flv_codecid) {
  81         case FLV_CODECID_H263  : vcodec->codec_id = CODEC_ID_FLV1   ; break;
  82         case FLV_CODECID_SCREEN: vcodec->codec_id = CODEC_ID_FLASHSV; break;
  83         case FLV_CODECID_VP6   : vcodec->codec_id = CODEC_ID_VP6F   ;
  84         case FLV_CODECID_VP6A  :
  85             if(flv_codecid == FLV_CODECID_VP6A)
  86                 vcodec->codec_id = CODEC_ID_VP6A;
  87             if(vcodec->extradata_size != 1) {
  88                 vcodec->extradata_size = 1;
  89                 vcodec->extradata = av_malloc(1);
  90             }
  91             vcodec->extradata[0] = get_byte(s->pb);
  92             return 1; // 1 byte body size adjustment for flv_read_packet()
  93         case FLV_CODECID_H264:
  94             vcodec->codec_id = CODEC_ID_H264;
  95             return 3; // not 4, reading packet type will consume one byte
  96         default:
  97             av_log(s, AV_LOG_INFO, "Unsupported video codec (%x)\n", flv_codecid);
  98             vcodec->codec_tag = flv_codecid;
  99     }
 100
 101     return 0;
 102 }
 103
 104 static int amf_get_string(ByteIOContext *ioc, char *buffer, int buffsize) {
 105     int length = get_be16(ioc);
 106     if(length >= buffsize) {
 107         url_fskip(ioc, length);
 108         return -1;
 109     }
 110
 111     get_buffer(ioc, buffer, length);
 112
 113     buffer[length] = '\0';
 114
 115     return length;
 116 }
 117
 118 static int amf_parse_object(AVFormatContext *s, AVStream *astream, AVStream *vstream, const char *key, int64_t max_pos, int depth) {
 119     AVCodecContext *acodec, *vcodec;
 120     ByteIOContext *ioc;
 121     AMFDataType amf_type;
 122     char str_val[256];
 123     double num_val;
 124
 125     num_val = 0;
 126     ioc = s->pb;
 127
 128     amf_type = get_byte(ioc);
 129
 130     switch(amf_type) {
 131         case AMF_DATA_TYPE_NUMBER:
 132             num_val = av_int2dbl(get_be64(ioc)); break;
 133         case AMF_DATA_TYPE_BOOL:
 134             num_val = get_byte(ioc); break;
 135         case AMF_DATA_TYPE_STRING:
 136             if(amf_get_string(ioc, str_val, sizeof(str_val)) < 0)
 137                 return -1;
 138             break;
 139         case AMF_DATA_TYPE_OBJECT: {
 140             unsigned int keylen;
 141
 142             while(url_ftell(ioc) < max_pos - 2 && (keylen = get_be16(ioc))) {
 143                 url_fskip(ioc, keylen); //skip key string
 144                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 145                     return -1; //if we couldn't skip, bomb out.
 146             }
 147             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 148                 return -1;
 149         }
 150             break;
 151         case AMF_DATA_TYPE_NULL:
 152         case AMF_DATA_TYPE_UNDEFINED:
 153         case AMF_DATA_TYPE_UNSUPPORTED:
 154             break; //these take up no additional space
 155         case AMF_DATA_TYPE_MIXEDARRAY:
 156             url_fskip(ioc, 4); //skip 32-bit max array index
 157             while(url_ftell(ioc) < max_pos - 2 && amf_get_string(ioc, str_val, sizeof(str_val)) > 0) {
 158                 //this is the only case in which we would want a nested parse to not skip over the object
 159                 if(amf_parse_object(s, astream, vstream, str_val, max_pos, depth + 1) < 0)
 160                     return -1;
 161             }
 162             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 163                 return -1;
 164             break;
 165         case AMF_DATA_TYPE_ARRAY: {
 166             unsigned int arraylen, i;
 167
 168             arraylen = get_be32(ioc);
 169             for(i = 0; i < arraylen && url_ftell(ioc) < max_pos - 1; i++) {
 170                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 171                     return -1; //if we couldn't skip, bomb out.
 172             }
 173         }
 174             break;
 175         case AMF_DATA_TYPE_DATE:
 176             url_fskip(ioc, 8 + 2); //timestamp (double) and UTC offset (int16)
 177             break;
 178         default: //unsupported type, we couldn't skip
 179             return -1;
 180     }
 181
 182     if(depth == 1 && key) { //only look for metadata values when we are not nested and key != NULL
 183         acodec = astream ? astream->codec : NULL;
 184         vcodec = vstream ? vstream->codec : NULL;
 185
 186         if(amf_type == AMF_DATA_TYPE_BOOL) {
 187             if(!strcmp(key, "stereo") && acodec) acodec->channels = num_val > 0 ? 2 : 1;
 188         } else if(amf_type == AMF_DATA_TYPE_NUMBER) {
 189             if(!strcmp(key, "duration")) s->duration = num_val * AV_TIME_BASE;
 190 //            else if(!strcmp(key, "width")  && vcodec && num_val > 0) vcodec->width  = num_val;
 191 //            else if(!strcmp(key, "height") && vcodec && num_val > 0) vcodec->height = num_val;
 192             else if(!strcmp(key, "videodatarate") && vcodec && 0 <= (int)(num_val * 1024.0))
 193                 vcodec->bit_rate = num_val * 1024.0;
 194             else if(!strcmp(key, "audiocodecid") && acodec && 0 <= (int)num_val)
 195                 flv_set_audio_codec(s, astream, (int)num_val << FLV_AUDIO_CODECID_OFFSET);
 196             else if(!strcmp(key, "videocodecid") && vcodec && 0 <= (int)num_val)
 197                 flv_set_video_codec(s, vstream, (int)num_val);
 198             else if(!strcmp(key, "audiosamplesize") && acodec && 0 < (int)num_val) {
 199                 acodec->bits_per_coded_sample = num_val;
 200                 //we may have to rewrite a previously read codecid because FLV only marks PCM endianness.
 201                 if(num_val == 8 && (acodec->codec_id == CODEC_ID_PCM_S16BE || acodec->codec_id == CODEC_ID_PCM_S16LE))
 202                     acodec->codec_id = CODEC_ID_PCM_S8;
 203             }
 204             else if(!strcmp(key, "audiosamplerate") && acodec && num_val >= 0) {
 205                 //some tools, like FLVTool2, write consistently approximate metadata sample rates
 206                 if (!acodec->sample_rate) {
 207                     switch((int)num_val) {
 208                         case 44000: acodec->sample_rate = 44100  ; break;
 209                         case 22000: acodec->sample_rate = 22050  ; break;
 210                         case 11000: acodec->sample_rate = 11025  ; break;
 211                         case 5000 : acodec->sample_rate = 5512   ; break;
 212                         default   : acodec->sample_rate = num_val;
 213                     }
 214                 }
 215             }
 216         }
 217     }
 218
 219     return 0;
 220 }
 221
 222 static int flv_read_metabody(AVFormatContext *s, int64_t next_pos) {
 223     AMFDataType type;
 224     AVStream *stream, *astream, *vstream;
 225     ByteIOContext *ioc;
 226     int i;
 227     char buffer[11]; //only needs to hold the string "onMetaData". Anything longer is something we don't want.
 228
 229     astream = NULL;
 230     vstream = NULL;
 231     ioc = s->pb;
 232
 233     //first object needs to be "onMetaData" string
 234     type = get_byte(ioc);
 235     if(type != AMF_DATA_TYPE_STRING || amf_get_string(ioc, buffer, sizeof(buffer)) < 0 || strcmp(buffer, "onMetaData"))
 236         return -1;
 237
 238     //find the streams now so that amf_parse_object doesn't need to do the lookup every time it is called.
 239     for(i = 0; i < s->nb_streams; i++) {
 240         stream = s->streams[i];
 241         if     (stream->codec->codec_type == CODEC_TYPE_AUDIO) astream = stream;
 242         else if(stream->codec->codec_type == CODEC_TYPE_VIDEO) vstream = stream;
 243     }
 244
 245     //parse the second object (we want a mixed array)
 246     if(amf_parse_object(s, astream, vstream, buffer, next_pos, 0) < 0)
 247         return -1;
 248
 249     return 0;
 250 }
 251
 252 static AVStream *create_stream(AVFormatContext *s, int is_audio){
 253     AVStream *st = av_new_stream(s, is_audio);
 254     if (!st)
 255         return NULL;
 256     st->codec->codec_type = is_audio ? CODEC_TYPE_AUDIO : CODEC_TYPE_VIDEO;
 257     av_set_pts_info(st, 32, 1, 1000); /* 32 bit pts in ms */
 258     return st;
 259 }
 260
 261 static int flv_read_header(AVFormatContext *s,
 262                            AVFormatParameters *ap)
 263 {
 264     int offset, flags;
 265
 266     url_fskip(s->pb, 4);
 267     flags = get_byte(s->pb);
 268     /* old flvtool cleared this field */
 269     /* FIXME: better fix needed */
 270     if (!flags) {
 271         flags = FLV_HEADER_FLAG_HASVIDEO | FLV_HEADER_FLAG_HASAUDIO;
 272         av_log(s, AV_LOG_WARNING, "Broken FLV file, which says no streams present, this might fail\n");
 273     }
 274
 275     if((flags & (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 276              != (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 277         s->ctx_flags |= AVFMTCTX_NOHEADER;
 278
 279     if(flags & FLV_HEADER_FLAG_HASVIDEO){
 280         if(!create_stream(s, 0))
 281             return AVERROR(ENOMEM);
 282     }
 283     if(flags & FLV_HEADER_FLAG_HASAUDIO){
 284         if(!create_stream(s, 1))
 285             return AVERROR(ENOMEM);
 286     }
 287
 288     offset = get_be32(s->pb);
 289     url_fseek(s->pb, offset, SEEK_SET);
 290
 291     s->start_time = 0;
 292
 293     return 0;
 294 }
 295
 296 static int flv_get_extradata(AVFormatContext *s, AVStream *st, int size)
 297 {
 298     av_free(st->codec->extradata);
 299     st->codec->extradata = av_mallocz(size + FF_INPUT_BUFFER_PADDING_SIZE);
 300     if (!st->codec->extradata)
 301         return AVERROR(ENOMEM);
 302     st->codec->extradata_size = size;
 303     get_buffer(s->pb, st->codec->extradata, st->codec->extradata_size);
 304     return 0;
 305 }
 306
 307 static int flv_read_packet(AVFormatContext *s, AVPacket *pkt)
 308 {
 309     FLVContext *flv = s->priv_data;
 310     int ret, i, type, size, flags, is_audio;
 311     int64_t next, pos;
 312     int64_t dts, pts = AV_NOPTS_VALUE;
 313     AVStream *st = NULL;
 314
 315  for(;;){
 316     pos = url_ftell(s->pb);
 317     url_fskip(s->pb, 4); /* size of previous packet */
 318     type = get_byte(s->pb);
 319     size = get_be24(s->pb);
 320     dts = get_be24(s->pb);
 321     dts |= get_byte(s->pb) << 24;
 322 //    av_log(s, AV_LOG_DEBUG, "type:%d, size:%d, dts:%d\n", type, size, dts);
 323     if (url_feof(s->pb))
 324         return AVERROR_EOF;
 325     url_fskip(s->pb, 3); /* stream id, always 0 */
 326     flags = 0;
 327
 328     if(size == 0)
 329         continue;
 330
 331     next= size + url_ftell(s->pb);
 332
 333     if (type == FLV_TAG_TYPE_AUDIO) {
 334         is_audio=1;
 335         flags = get_byte(s->pb);
 336         size--;
 337     } else if (type == FLV_TAG_TYPE_VIDEO) {
 338         is_audio=0;
 339         flags = get_byte(s->pb);
 340         size--;
 341         if ((flags & 0xf0) == 0x50) /* video info / command frame */
 342             goto skip;
 343     } else {
 344         if (type == FLV_TAG_TYPE_META && size > 13+1+4 && 0)
 345             flv_read_metabody(s, next);
 346         else /* skip packet */
 347             av_log(s, AV_LOG_ERROR, "skipping flv packet: type %d, size %d, flags %d\n", type, size, flags);
 348     skip:
 349         url_fseek(s->pb, next, SEEK_SET);
 350         continue;
 351     }
 352
 353     /* skip empty data packets */
 354     if (!size)
 355         continue;
 356
 357     /* now find stream */
 358     for(i=0;i<s->nb_streams;i++) {
 359         st = s->streams[i];
 360         if (st->id == is_audio)
 361             break;
 362     }
 363     if(i == s->nb_streams){
 364         av_log(s, AV_LOG_ERROR, "invalid stream\n");
 365         st= create_stream(s, is_audio);
 366         s->ctx_flags &= ~AVFMTCTX_NOHEADER;
 367     }
 368 //    av_log(s, AV_LOG_DEBUG, "%d %X %d \n", is_audio, flags, st->discard);
 369     if(  (st->discard >= AVDISCARD_NONKEY && !((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY ||         is_audio))
 370        ||(st->discard >= AVDISCARD_BIDIR  &&  ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_DISP_INTER && !is_audio))
 371        || st->discard >= AVDISCARD_ALL
 372        ){
 373         url_fseek(s->pb, next, SEEK_SET);
 374         continue;
 375     }
 376     if ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY)
 377         av_add_index_entry(st, pos, dts, size, 0, AVINDEX_KEYFRAME);
 378     break;
 379  }
 380
 381     // if not streamed and no duration from metadata then seek to end to find the duration from the timestamps
 382     if(!url_is_streamed(s->pb) && s->duration==AV_NOPTS_VALUE){
 383         int size;
 384         const int64_t pos= url_ftell(s->pb);
 385         const int64_t fsize= url_fsize(s->pb);
 386         url_fseek(s->pb, fsize-4, SEEK_SET);
 387         size= get_be32(s->pb);
 388         url_fseek(s->pb, fsize-3-size, SEEK_SET);
 389         if(size == get_be24(s->pb) + 11){
 390             s->duration= get_be24(s->pb) * (int64_t)AV_TIME_BASE / 1000;
 391         }
 392         url_fseek(s->pb, pos, SEEK_SET);
 393     }
 394
 395     if(is_audio){
 396         if(!st->codec->channels || !st->codec->sample_rate || !st->codec->bits_per_coded_sample) {
 397             st->codec->channels = (flags & FLV_AUDIO_CHANNEL_MASK) == FLV_STEREO ? 2 : 1;
 398             st->codec->sample_rate = (44100 << ((flags & FLV_AUDIO_SAMPLERATE_MASK) >> FLV_AUDIO_SAMPLERATE_OFFSET) >> 3);
 399             st->codec->bits_per_coded_sample = (flags & FLV_AUDIO_SAMPLESIZE_MASK) ? 16 : 8;
 400         }
 401         if(!st->codec->codec_id){
 402             flv_set_audio_codec(s, st, flags & FLV_AUDIO_CODECID_MASK);
 403         }
 404     }else{
 405         size -= flv_set_video_codec(s, st, flags & FLV_VIDEO_CODECID_MASK);
 406     }
 407
 408     if (st->codec->codec_id == CODEC_ID_AAC ||
 409         st->codec->codec_id == CODEC_ID_H264) {
 410         int type = get_byte(s->pb);
 411         size--;
 412         if (st->codec->codec_id == CODEC_ID_H264) {
 413             int32_t cts = (get_be24(s->pb)+0xff800000)^0xff800000; // sign extension
 414             pts = dts + cts;
 415             if (cts < 0) { // dts are wrong
 416                 flv->wrong_dts = 1;
 417                 av_log(s, AV_LOG_WARNING, "negative cts, previous timestamps might be wrong\n");
 418             }
 419             if (flv->wrong_dts)
 420                 dts = AV_NOPTS_VALUE;
 421         }
 422         if (type == 0) {
 423             if ((ret = flv_get_extradata(s, st, size)) < 0)
 424                 return ret;
 425             if (st->codec->codec_id == CODEC_ID_AAC) {
 426                 MPEG4AudioConfig cfg;
 427                 ff_mpeg4audio_get_config(&cfg, st->codec->extradata,
 428                                          st->codec->extradata_size);
 429                 if (cfg.chan_config > 7)
 430                     return -1;
 431                 st->codec->channels = ff_mpeg4audio_channels[cfg.chan_config];
 432                 st->codec->sample_rate = cfg.sample_rate;
 433                 dprintf(s, "mp4a config channels %d sample rate %d\n",
 434                         st->codec->channels, st->codec->sample_rate);
 435             }
 436
 437             return AVERROR(EAGAIN);
 438         }
 439     }
 440
 441     /* skip empty data packets */
 442     if (!size)
 443         return AVERROR(EAGAIN);
 444
 445     ret= av_get_packet(s->pb, pkt, size);
 446     if (ret < 0) {
 447         return AVERROR(EIO);
 448     }
 449     /* note: we need to modify the packet size here to handle the last
 450        packet */
 451     pkt->size = ret;
 452     pkt->dts = dts;
 453     pkt->pts = pts == AV_NOPTS_VALUE ? dts : pts;
 454     pkt->stream_index = st->index;
 455
 456     if (is_audio || ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY))
 457         pkt->flags |= PKT_FLAG_KEY;
 458
 459     return ret;
 460 }
 461
 462 AVInputFormat flv_demuxer = {
 463     "flv",
 464     NULL_IF_CONFIG_SMALL("FLV format"),
 465     sizeof(FLVContext),
 466     flv_probe,
 467     flv_read_header,
 468     flv_read_packet,
 469     .extensions = "flv",
 470     .value = CODEC_ID_FLV1,
 471 };