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_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_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: acodec->codec_id = CODEC_ID_SPEEX;                                  break;
  57         case FLV_CODECID_MP3  : acodec->codec_id = CODEC_ID_MP3      ; astream->need_parsing = AVSTREAM_PARSE_FULL; break;
  58         case FLV_CODECID_NELLYMOSER_8HZ_MONO:
  59             acodec->sample_rate = 8000; //in case metadata does not otherwise declare samplerate
  60         case FLV_CODECID_NELLYMOSER:
  61             acodec->codec_id = CODEC_ID_NELLYMOSER;
  62             break;
  63         default:
  64             av_log(s, AV_LOG_INFO, "Unsupported audio codec (%x)\n", flv_codecid >> FLV_AUDIO_CODECID_OFFSET);
  65             acodec->codec_tag = flv_codecid >> FLV_AUDIO_CODECID_OFFSET;
  66     }
  67 }
  68
  69 static int flv_set_video_codec(AVFormatContext *s, AVStream *vstream, int flv_codecid) {
  70     AVCodecContext *vcodec = vstream->codec;
  71     switch(flv_codecid) {
  72         case FLV_CODECID_H263  : vcodec->codec_id = CODEC_ID_FLV1   ; break;
  73         case FLV_CODECID_SCREEN: vcodec->codec_id = CODEC_ID_FLASHSV; break;
  74         case FLV_CODECID_VP6   : vcodec->codec_id = CODEC_ID_VP6F   ;
  75         case FLV_CODECID_VP6A  :
  76             if(flv_codecid == FLV_CODECID_VP6A)
  77                 vcodec->codec_id = CODEC_ID_VP6A;
  78             if(vcodec->extradata_size != 1) {
  79                 vcodec->extradata_size = 1;
  80                 vcodec->extradata = av_malloc(1);
  81             }
  82             vcodec->extradata[0] = get_byte(s->pb);
  83             return 1; // 1 byte body size adjustment for flv_read_packet()
  84         case FLV_CODECID_H264:
  85             vcodec->codec_id = CODEC_ID_H264;
  86             return 3; // not 4, reading packet type will consume one byte
  87         default:
  88             av_log(s, AV_LOG_INFO, "Unsupported video codec (%x)\n", flv_codecid);
  89             vcodec->codec_tag = flv_codecid;
  90     }
  91
  92     return 0;
  93 }
  94
  95 static int amf_get_string(ByteIOContext *ioc, char *buffer, int buffsize) {
  96     int length = get_be16(ioc);
  97     if(length >= buffsize) {
  98         url_fskip(ioc, length);
  99         return -1;
 100     }
 101
 102     get_buffer(ioc, buffer, length);
 103
 104     buffer[length] = '\0';
 105
 106     return length;
 107 }
 108
 109 static int amf_parse_object(AVFormatContext *s, AVStream *astream, AVStream *vstream, const char *key, unsigned int max_pos, int depth) {
 110     AVCodecContext *acodec, *vcodec;
 111     ByteIOContext *ioc;
 112     AMFDataType amf_type;
 113     char str_val[256];
 114     double num_val;
 115
 116     num_val = 0;
 117     ioc = s->pb;
 118
 119     amf_type = get_byte(ioc);
 120
 121     switch(amf_type) {
 122         case AMF_DATA_TYPE_NUMBER:
 123             num_val = av_int2dbl(get_be64(ioc)); break;
 124         case AMF_DATA_TYPE_BOOL:
 125             num_val = get_byte(ioc); break;
 126         case AMF_DATA_TYPE_STRING:
 127             if(amf_get_string(ioc, str_val, sizeof(str_val)) < 0)
 128                 return -1;
 129             break;
 130         case AMF_DATA_TYPE_OBJECT: {
 131             unsigned int keylen;
 132
 133             while(url_ftell(ioc) < max_pos - 2 && (keylen = get_be16(ioc))) {
 134                 url_fskip(ioc, keylen); //skip key string
 135                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 136                     return -1; //if we couldn't skip, bomb out.
 137             }
 138             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 139                 return -1;
 140         }
 141             break;
 142         case AMF_DATA_TYPE_NULL:
 143         case AMF_DATA_TYPE_UNDEFINED:
 144         case AMF_DATA_TYPE_UNSUPPORTED:
 145             break; //these take up no additional space
 146         case AMF_DATA_TYPE_MIXEDARRAY:
 147             url_fskip(ioc, 4); //skip 32-bit max array index
 148             while(url_ftell(ioc) < max_pos - 2 && amf_get_string(ioc, str_val, sizeof(str_val)) > 0) {
 149                 //this is the only case in which we would want a nested parse to not skip over the object
 150                 if(amf_parse_object(s, astream, vstream, str_val, max_pos, depth + 1) < 0)
 151                     return -1;
 152             }
 153             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 154                 return -1;
 155             break;
 156         case AMF_DATA_TYPE_ARRAY: {
 157             unsigned int arraylen, i;
 158
 159             arraylen = get_be32(ioc);
 160             for(i = 0; i < arraylen && url_ftell(ioc) < max_pos - 1; i++) {
 161                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 162                     return -1; //if we couldn't skip, bomb out.
 163             }
 164         }
 165             break;
 166         case AMF_DATA_TYPE_DATE:
 167             url_fskip(ioc, 8 + 2); //timestamp (double) and UTC offset (int16)
 168             break;
 169         default: //unsupported type, we couldn't skip
 170             return -1;
 171     }
 172
 173     if(depth == 1 && key) { //only look for metadata values when we are not nested and key != NULL
 174         acodec = astream ? astream->codec : NULL;
 175         vcodec = vstream ? vstream->codec : NULL;
 176
 177         if(amf_type == AMF_DATA_TYPE_BOOL) {
 178             if(!strcmp(key, "stereo") && acodec) acodec->channels = num_val > 0 ? 2 : 1;
 179         } else if(amf_type == AMF_DATA_TYPE_NUMBER) {
 180             if(!strcmp(key, "duration")) s->duration = num_val * AV_TIME_BASE;
 181 //            else if(!strcmp(key, "width")  && vcodec && num_val > 0) vcodec->width  = num_val;
 182 //            else if(!strcmp(key, "height") && vcodec && num_val > 0) vcodec->height = num_val;
 183             else if(!strcmp(key, "audiocodecid") && acodec && 0 <= (int)num_val)
 184                 flv_set_audio_codec(s, astream, (int)num_val << FLV_AUDIO_CODECID_OFFSET);
 185             else if(!strcmp(key, "videocodecid") && vcodec && 0 <= (int)num_val)
 186                 flv_set_video_codec(s, vstream, (int)num_val);
 187             else if(!strcmp(key, "audiosamplesize") && acodec && 0 < (int)num_val) {
 188                 acodec->bits_per_sample = num_val;
 189                 //we may have to rewrite a previously read codecid because FLV only marks PCM endianness.
 190                 if(num_val == 8 && (acodec->codec_id == CODEC_ID_PCM_S16BE || acodec->codec_id == CODEC_ID_PCM_S16LE))
 191                     acodec->codec_id = CODEC_ID_PCM_S8;
 192             }
 193             else if(!strcmp(key, "audiosamplerate") && acodec && num_val >= 0) {
 194                 //some tools, like FLVTool2, write consistently approximate metadata sample rates
 195                 if (!acodec->sample_rate) {
 196                     switch((int)num_val) {
 197                         case 44000: acodec->sample_rate = 44100  ; break;
 198                         case 22000: acodec->sample_rate = 22050  ; break;
 199                         case 11000: acodec->sample_rate = 11025  ; break;
 200                         case 5000 : acodec->sample_rate = 5512   ; break;
 201                         default   : acodec->sample_rate = num_val;
 202                     }
 203                 }
 204             }
 205         }
 206     }
 207
 208     return 0;
 209 }
 210
 211 static int flv_read_metabody(AVFormatContext *s, unsigned int next_pos) {
 212     AMFDataType type;
 213     AVStream *stream, *astream, *vstream;
 214     ByteIOContext *ioc;
 215     int i, keylen;
 216     char buffer[11]; //only needs to hold the string "onMetaData". Anything longer is something we don't want.
 217
 218     astream = NULL;
 219     vstream = NULL;
 220     keylen = 0;
 221     ioc = s->pb;
 222
 223     //first object needs to be "onMetaData" string
 224     type = get_byte(ioc);
 225     if(type != AMF_DATA_TYPE_STRING || amf_get_string(ioc, buffer, sizeof(buffer)) < 0 || strcmp(buffer, "onMetaData"))
 226         return -1;
 227
 228     //find the streams now so that amf_parse_object doesn't need to do the lookup every time it is called.
 229     for(i = 0; i < s->nb_streams; i++) {
 230         stream = s->streams[i];
 231         if     (stream->codec->codec_type == CODEC_TYPE_AUDIO) astream = stream;
 232         else if(stream->codec->codec_type == CODEC_TYPE_VIDEO) vstream = stream;
 233     }
 234
 235     //parse the second object (we want a mixed array)
 236     if(amf_parse_object(s, astream, vstream, buffer, next_pos, 0) < 0)
 237         return -1;
 238
 239     return 0;
 240 }
 241
 242 static AVStream *create_stream(AVFormatContext *s, int is_audio){
 243     AVStream *st = av_new_stream(s, is_audio);
 244     if (!st)
 245         return NULL;
 246     st->codec->codec_type = is_audio ? CODEC_TYPE_AUDIO : CODEC_TYPE_VIDEO;
 247     av_set_pts_info(st, 32, 1, 1000); /* 32 bit pts in ms */
 248     return st;
 249 }
 250
 251 static int flv_read_header(AVFormatContext *s,
 252                            AVFormatParameters *ap)
 253 {
 254     int offset, flags;
 255
 256     url_fskip(s->pb, 4);
 257     flags = get_byte(s->pb);
 258     /* old flvtool cleared this field */
 259     /* FIXME: better fix needed */
 260     if (!flags) {
 261         flags = FLV_HEADER_FLAG_HASVIDEO | FLV_HEADER_FLAG_HASAUDIO;
 262         av_log(s, AV_LOG_WARNING, "Broken FLV file, which says no streams present, this might fail\n");
 263     }
 264
 265     if((flags & (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 266              != (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 267         s->ctx_flags |= AVFMTCTX_NOHEADER;
 268
 269     if(flags & FLV_HEADER_FLAG_HASVIDEO){
 270         if(!create_stream(s, 0))
 271             return AVERROR(ENOMEM);
 272     }
 273     if(flags & FLV_HEADER_FLAG_HASAUDIO){
 274         if(!create_stream(s, 1))
 275             return AVERROR(ENOMEM);
 276     }
 277
 278     offset = get_be32(s->pb);
 279     url_fseek(s->pb, offset, SEEK_SET);
 280
 281     s->start_time = 0;
 282
 283     return 0;
 284 }
 285
 286 static int flv_get_extradata(AVFormatContext *s, AVStream *st, int size)
 287 {
 288     av_free(st->codec->extradata);
 289     st->codec->extradata = av_mallocz(size + FF_INPUT_BUFFER_PADDING_SIZE);
 290     if (!st->codec->extradata)
 291         return AVERROR(ENOMEM);
 292     st->codec->extradata_size = size;
 293     get_buffer(s->pb, st->codec->extradata, st->codec->extradata_size);
 294     return 0;
 295 }
 296
 297 static int flv_read_packet(AVFormatContext *s, AVPacket *pkt)
 298 {
 299     int ret, i, type, size, flags, is_audio, next, pos;
 300     unsigned dts;
 301     AVStream *st = NULL;
 302
 303  retry:
 304  for(;;){
 305     pos = url_ftell(s->pb);
 306     url_fskip(s->pb, 4); /* size of previous packet */
 307     type = get_byte(s->pb);
 308     size = get_be24(s->pb);
 309     dts = get_be24(s->pb);
 310     dts |= get_byte(s->pb) << 24;
 311 //    av_log(s, AV_LOG_DEBUG, "type:%d, size:%d, dts:%d\n", type, size, dts);
 312     if (url_feof(s->pb))
 313         return AVERROR(EIO);
 314     url_fskip(s->pb, 3); /* stream id, always 0 */
 315     flags = 0;
 316
 317     if(size == 0)
 318         continue;
 319
 320     next= size + url_ftell(s->pb);
 321
 322     if (type == FLV_TAG_TYPE_AUDIO) {
 323         is_audio=1;
 324         flags = get_byte(s->pb);
 325         size--;
 326     } else if (type == FLV_TAG_TYPE_VIDEO) {
 327         is_audio=0;
 328         flags = get_byte(s->pb);
 329         size--;
 330         if ((flags & 0xf0) == 0x50) /* video info / command frame */
 331             goto skip;
 332     } else {
 333         if (type == FLV_TAG_TYPE_META && size > 13+1+4)
 334             flv_read_metabody(s, next);
 335         else /* skip packet */
 336             av_log(s, AV_LOG_ERROR, "skipping flv packet: type %d, size %d, flags %d\n", type, size, flags);
 337     skip:
 338         url_fseek(s->pb, next, SEEK_SET);
 339         continue;
 340     }
 341
 342     /* skip empty data packets */
 343     if (!size)
 344         continue;
 345
 346     /* now find stream */
 347     for(i=0;i<s->nb_streams;i++) {
 348         st = s->streams[i];
 349         if (st->id == is_audio)
 350             break;
 351     }
 352     if(i == s->nb_streams){
 353         av_log(NULL, AV_LOG_ERROR, "invalid stream\n");
 354         st= create_stream(s, is_audio);
 355         s->ctx_flags &= ~AVFMTCTX_NOHEADER;
 356     }
 357 //    av_log(NULL, AV_LOG_DEBUG, "%d %X %d \n", is_audio, flags, st->discard);
 358     if(  (st->discard >= AVDISCARD_NONKEY && !((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY ||         is_audio))
 359        ||(st->discard >= AVDISCARD_BIDIR  &&  ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_DISP_INTER && !is_audio))
 360        || st->discard >= AVDISCARD_ALL
 361        ){
 362         url_fseek(s->pb, next, SEEK_SET);
 363         continue;
 364     }
 365     if ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY)
 366         av_add_index_entry(st, pos, dts, size, 0, AVINDEX_KEYFRAME);
 367     break;
 368  }
 369
 370     // if not streamed and no duration from metadata then seek to end to find the duration from the timestamps
 371     if(!url_is_streamed(s->pb) && s->duration==AV_NOPTS_VALUE){
 372         int size;
 373         const int pos= url_ftell(s->pb);
 374         const int fsize= url_fsize(s->pb);
 375         url_fseek(s->pb, fsize-4, SEEK_SET);
 376         size= get_be32(s->pb);
 377         url_fseek(s->pb, fsize-3-size, SEEK_SET);
 378         if(size == get_be24(s->pb) + 11){
 379             s->duration= get_be24(s->pb) * (int64_t)AV_TIME_BASE / 1000;
 380         }
 381         url_fseek(s->pb, pos, SEEK_SET);
 382     }
 383
 384     if(is_audio){
 385         if(!st->codec->sample_rate || !st->codec->bits_per_sample || (!st->codec->codec_id && !st->codec->codec_tag)) {
 386             st->codec->channels = (flags & FLV_AUDIO_CHANNEL_MASK) == FLV_STEREO ? 2 : 1;
 387             if((flags & FLV_AUDIO_CODECID_MASK) == FLV_CODECID_NELLYMOSER_8HZ_MONO)
 388                 st->codec->sample_rate= 8000;
 389             else
 390                 st->codec->sample_rate = (44100 << ((flags & FLV_AUDIO_SAMPLERATE_MASK) >> FLV_AUDIO_SAMPLERATE_OFFSET) >> 3);
 391             st->codec->bits_per_sample = (flags & FLV_AUDIO_SAMPLESIZE_MASK) ? 16 : 8;
 392             flv_set_audio_codec(s, st, flags & FLV_AUDIO_CODECID_MASK);
 393         }
 394     }else{
 395         size -= flv_set_video_codec(s, st, flags & FLV_VIDEO_CODECID_MASK);
 396     }
 397
 398     if (st->codec->codec_id == CODEC_ID_AAC ||
 399         st->codec->codec_id == CODEC_ID_H264) {
 400         int type = get_byte(s->pb);
 401         size--;
 402         if (st->codec->codec_id == CODEC_ID_H264) {
 403             // cts offset ignored because it might to be signed
 404             // and would cause pts < dts
 405             get_be24(s->pb);
 406         }
 407         if (type == 0) {
 408             if ((ret = flv_get_extradata(s, st, size)) < 0)
 409                 return ret;
 410             goto retry;
 411         }
 412     }
 413
 414     ret= av_get_packet(s->pb, pkt, size);
 415     if (ret <= 0) {
 416         return AVERROR(EIO);
 417     }
 418     /* note: we need to modify the packet size here to handle the last
 419        packet */
 420     pkt->size = ret;
 421     pkt->dts = dts;
 422     pkt->stream_index = st->index;
 423
 424     if (is_audio || ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY))
 425         pkt->flags |= PKT_FLAG_KEY;
 426
 427     return ret;
 428 }
 429
 430 AVInputFormat flv_demuxer = {
 431     "flv",
 432     NULL_IF_CONFIG_SMALL("FLV format"),
 433     0,
 434     flv_probe,
 435     flv_read_header,
 436     flv_read_packet,
 437     .extensions = "flv",
 438     .value = CODEC_ID_FLV1,
 439 };