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