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_MP3  : acodec->codec_id = CODEC_ID_MP3      ; astream->need_parsing = AVSTREAM_PARSE_FULL; break;
  57         case FLV_CODECID_NELLYMOSER_8HZ_MONO:
  58             acodec->sample_rate = 8000; //in case metadata does not otherwise declare samplerate
  59         case FLV_CODECID_NELLYMOSER:
  60             acodec->codec_id = CODEC_ID_NELLYMOSER;
  61             break;
  62         default:
  63             av_log(s, AV_LOG_INFO, "Unsupported audio codec (%x)\n", flv_codecid >> FLV_AUDIO_CODECID_OFFSET);
  64             acodec->codec_tag = flv_codecid >> FLV_AUDIO_CODECID_OFFSET;
  65     }
  66 }
  67
  68 static int flv_set_video_codec(AVFormatContext *s, AVStream *vstream, int flv_codecid) {
  69     AVCodecContext *vcodec = vstream->codec;
  70     switch(flv_codecid) {
  71         case FLV_CODECID_H263  : vcodec->codec_id = CODEC_ID_FLV1   ; break;
  72         case FLV_CODECID_SCREEN: vcodec->codec_id = CODEC_ID_FLASHSV; break;
  73         case FLV_CODECID_VP6   : vcodec->codec_id = CODEC_ID_VP6F   ;
  74         case FLV_CODECID_VP6A  :
  75             if(flv_codecid == FLV_CODECID_VP6A)
  76                 vcodec->codec_id = CODEC_ID_VP6A;
  77             if(vcodec->extradata_size != 1) {
  78                 vcodec->extradata_size = 1;
  79                 vcodec->extradata = av_malloc(1);
  80             }
  81             vcodec->extradata[0] = get_byte(s->pb);
  82             return 1; // 1 byte body size adjustment for flv_read_packet()
  83         case FLV_CODECID_H264:
  84             vcodec->codec_id = CODEC_ID_H264;
  85             return 3; // not 4, reading packet type will consume one byte
  86         default:
  87             av_log(s, AV_LOG_INFO, "Unsupported video codec (%x)\n", flv_codecid);
  88             vcodec->codec_tag = flv_codecid;
  89     }
  90
  91     return 0;
  92 }
  93
  94 static int amf_get_string(ByteIOContext *ioc, char *buffer, int buffsize) {
  95     int length = get_be16(ioc);
  96     if(length >= buffsize) {
  97         url_fskip(ioc, length);
  98         return -1;
  99     }
 100
 101     get_buffer(ioc, buffer, length);
 102
 103     buffer[length] = '\0';
 104
 105     return length;
 106 }
 107
 108 static int amf_parse_object(AVFormatContext *s, AVStream *astream, AVStream *vstream, const char *key, unsigned int max_pos, int depth) {
 109     AVCodecContext *acodec, *vcodec;
 110     ByteIOContext *ioc;
 111     AMFDataType amf_type;
 112     char str_val[256];
 113     double num_val;
 114
 115     num_val = 0;
 116     ioc = s->pb;
 117
 118     amf_type = get_byte(ioc);
 119
 120     switch(amf_type) {
 121         case AMF_DATA_TYPE_NUMBER:
 122             num_val = av_int2dbl(get_be64(ioc)); break;
 123         case AMF_DATA_TYPE_BOOL:
 124             num_val = get_byte(ioc); break;
 125         case AMF_DATA_TYPE_STRING:
 126             if(amf_get_string(ioc, str_val, sizeof(str_val)) < 0)
 127                 return -1;
 128             break;
 129         case AMF_DATA_TYPE_OBJECT: {
 130             unsigned int keylen;
 131
 132             while(url_ftell(ioc) < max_pos - 2 && (keylen = get_be16(ioc))) {
 133                 url_fskip(ioc, keylen); //skip key string
 134                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 135                     return -1; //if we couldn't skip, bomb out.
 136             }
 137             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 138                 return -1;
 139         }
 140             break;
 141         case AMF_DATA_TYPE_NULL:
 142         case AMF_DATA_TYPE_UNDEFINED:
 143         case AMF_DATA_TYPE_UNSUPPORTED:
 144             break; //these take up no additional space
 145         case AMF_DATA_TYPE_MIXEDARRAY:
 146             url_fskip(ioc, 4); //skip 32-bit max array index
 147             while(url_ftell(ioc) < max_pos - 2 && amf_get_string(ioc, str_val, sizeof(str_val)) > 0) {
 148                 //this is the only case in which we would want a nested parse to not skip over the object
 149                 if(amf_parse_object(s, astream, vstream, str_val, max_pos, depth + 1) < 0)
 150                     return -1;
 151             }
 152             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 153                 return -1;
 154             break;
 155         case AMF_DATA_TYPE_ARRAY: {
 156             unsigned int arraylen, i;
 157
 158             arraylen = get_be32(ioc);
 159             for(i = 0; i < arraylen && url_ftell(ioc) < max_pos - 1; i++) {
 160                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 161                     return -1; //if we couldn't skip, bomb out.
 162             }
 163         }
 164             break;
 165         case AMF_DATA_TYPE_DATE:
 166             url_fskip(ioc, 8 + 2); //timestamp (double) and UTC offset (int16)
 167             break;
 168         default: //unsupported type, we couldn't skip
 169             return -1;
 170     }
 171
 172     if(depth == 1 && key) { //only look for metadata values when we are not nested and key != NULL
 173         acodec = astream ? astream->codec : NULL;
 174         vcodec = vstream ? vstream->codec : NULL;
 175
 176         if(amf_type == AMF_DATA_TYPE_BOOL) {
 177             if(!strcmp(key, "stereo") && acodec) acodec->channels = num_val > 0 ? 2 : 1;
 178         } else if(amf_type == AMF_DATA_TYPE_NUMBER) {
 179             if(!strcmp(key, "duration")) s->duration = num_val * AV_TIME_BASE;
 180 //            else if(!strcmp(key, "width")  && vcodec && num_val > 0) vcodec->width  = num_val;
 181 //            else if(!strcmp(key, "height") && vcodec && num_val > 0) vcodec->height = num_val;
 182             else if(!strcmp(key, "audiocodecid") && acodec && 0 <= (int)num_val)
 183                 flv_set_audio_codec(s, astream, (int)num_val << FLV_AUDIO_CODECID_OFFSET);
 184             else if(!strcmp(key, "videocodecid") && vcodec && 0 <= (int)num_val)
 185                 flv_set_video_codec(s, vstream, (int)num_val);
 186             else if(!strcmp(key, "audiosamplesize") && acodec && 0 < (int)num_val) {
 187                 acodec->bits_per_sample = num_val;
 188                 //we may have to rewrite a previously read codecid because FLV only marks PCM endianness.
 189                 if(num_val == 8 && (acodec->codec_id == CODEC_ID_PCM_S16BE || acodec->codec_id == CODEC_ID_PCM_S16LE))
 190                     acodec->codec_id = CODEC_ID_PCM_S8;
 191             }
 192             else if(!strcmp(key, "audiosamplerate") && acodec && num_val >= 0) {
 193                 //some tools, like FLVTool2, write consistently approximate metadata sample rates
 194                 if (!acodec->sample_rate) {
 195                     switch((int)num_val) {
 196                         case 44000: acodec->sample_rate = 44100  ; break;
 197                         case 22000: acodec->sample_rate = 22050  ; break;
 198                         case 11000: acodec->sample_rate = 11025  ; break;
 199                         case 5000 : acodec->sample_rate = 5512   ; break;
 200                         default   : acodec->sample_rate = num_val;
 201                     }
 202                 }
 203             }
 204         }
 205     }
 206
 207     return 0;
 208 }
 209
 210 static int flv_read_metabody(AVFormatContext *s, unsigned int next_pos) {
 211     AMFDataType type;
 212     AVStream *stream, *astream, *vstream;
 213     ByteIOContext *ioc;
 214     int i, keylen;
 215     char buffer[11]; //only needs to hold the string "onMetaData". Anything longer is something we don't want.
 216
 217     astream = NULL;
 218     vstream = NULL;
 219     keylen = 0;
 220     ioc = s->pb;
 221
 222     //first object needs to be "onMetaData" string
 223     type = get_byte(ioc);
 224     if(type != AMF_DATA_TYPE_STRING || amf_get_string(ioc, buffer, sizeof(buffer)) < 0 || strcmp(buffer, "onMetaData"))
 225         return -1;
 226
 227     //find the streams now so that amf_parse_object doesn't need to do the lookup every time it is called.
 228     for(i = 0; i < s->nb_streams; i++) {
 229         stream = s->streams[i];
 230         if     (stream->codec->codec_type == CODEC_TYPE_AUDIO) astream = stream;
 231         else if(stream->codec->codec_type == CODEC_TYPE_VIDEO) vstream = stream;
 232     }
 233
 234     //parse the second object (we want a mixed array)
 235     if(amf_parse_object(s, astream, vstream, buffer, next_pos, 0) < 0)
 236         return -1;
 237
 238     return 0;
 239 }
 240
 241 static AVStream *create_stream(AVFormatContext *s, int is_audio){
 242     AVStream *st = av_new_stream(s, is_audio);
 243     if (!st)
 244         return NULL;
 245     st->codec->codec_type = is_audio ? CODEC_TYPE_AUDIO : CODEC_TYPE_VIDEO;
 246     av_set_pts_info(st, 32, 1, 1000); /* 32 bit pts in ms */
 247     return st;
 248 }
 249
 250 static int flv_read_header(AVFormatContext *s,
 251                            AVFormatParameters *ap)
 252 {
 253     int offset, flags;
 254
 255     url_fskip(s->pb, 4);
 256     flags = get_byte(s->pb);
 257     /* old flvtool cleared this field */
 258     /* FIXME: better fix needed */
 259     if (!flags) {
 260         flags = FLV_HEADER_FLAG_HASVIDEO | FLV_HEADER_FLAG_HASAUDIO;
 261         av_log(s, AV_LOG_WARNING, "Broken FLV file, which says no streams present, this might fail\n");
 262     }
 263
 264     if((flags & (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 265              != (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 266         s->ctx_flags |= AVFMTCTX_NOHEADER;
 267
 268     if(flags & FLV_HEADER_FLAG_HASVIDEO){
 269         if(!create_stream(s, 0))
 270             return AVERROR(ENOMEM);
 271     }
 272     if(flags & FLV_HEADER_FLAG_HASAUDIO){
 273         if(!create_stream(s, 1))
 274             return AVERROR(ENOMEM);
 275     }
 276
 277     offset = get_be32(s->pb);
 278     url_fseek(s->pb, offset, SEEK_SET);
 279
 280     s->start_time = 0;
 281
 282     return 0;
 283 }
 284
 285 static int flv_get_extradata(AVFormatContext *s, AVStream *st, int size)
 286 {
 287     av_free(st->codec->extradata);
 288     st->codec->extradata = av_mallocz(size + FF_INPUT_BUFFER_PADDING_SIZE);
 289     if (!st->codec->extradata)
 290         return AVERROR(ENOMEM);
 291     st->codec->extradata_size = size;
 292     get_buffer(s->pb, st->codec->extradata, st->codec->extradata_size);
 293     return 0;
 294 }
 295
 296 static int flv_read_packet(AVFormatContext *s, AVPacket *pkt)
 297 {
 298     int ret, i, type, size, flags, is_audio, next, pos;
 299     unsigned dts;
 300     AVStream *st = NULL;
 301
 302  retry:
 303  for(;;){
 304     pos = url_ftell(s->pb);
 305     url_fskip(s->pb, 4); /* size of previous packet */
 306     type = get_byte(s->pb);
 307     size = get_be24(s->pb);
 308     dts = get_be24(s->pb);
 309     dts |= get_byte(s->pb) << 24;
 310 //    av_log(s, AV_LOG_DEBUG, "type:%d, size:%d, dts:%d\n", type, size, dts);
 311     if (url_feof(s->pb))
 312         return AVERROR(EIO);
 313     url_fskip(s->pb, 3); /* stream id, always 0 */
 314     flags = 0;
 315
 316     if(size == 0)
 317         continue;
 318
 319     next= size + url_ftell(s->pb);
 320
 321     if (type == FLV_TAG_TYPE_AUDIO) {
 322         is_audio=1;
 323         flags = get_byte(s->pb);
 324     } else if (type == FLV_TAG_TYPE_VIDEO) {
 325         is_audio=0;
 326         flags = get_byte(s->pb);
 327     } else {
 328         if (type == FLV_TAG_TYPE_META && size > 13+1+4)
 329             flv_read_metabody(s, next);
 330         else /* skip packet */
 331             av_log(s, AV_LOG_ERROR, "skipping flv packet: type %d, size %d, flags %d\n", type, size, flags);
 332         url_fseek(s->pb, next, SEEK_SET);
 333         continue;
 334     }
 335
 336     /* now find stream */
 337     for(i=0;i<s->nb_streams;i++) {
 338         st = s->streams[i];
 339         if (st->id == is_audio)
 340             break;
 341     }
 342     if(i == s->nb_streams){
 343         av_log(NULL, AV_LOG_ERROR, "invalid stream\n");
 344         st= create_stream(s, is_audio);
 345         s->ctx_flags &= ~AVFMTCTX_NOHEADER;
 346     }
 347 //    av_log(NULL, AV_LOG_DEBUG, "%d %X %d \n", is_audio, flags, st->discard);
 348     if(  (st->discard >= AVDISCARD_NONKEY && !((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY ||         is_audio))
 349        ||(st->discard >= AVDISCARD_BIDIR  &&  ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_DISP_INTER && !is_audio))
 350        || st->discard >= AVDISCARD_ALL
 351        ){
 352         url_fseek(s->pb, next, SEEK_SET);
 353         continue;
 354     }
 355     if ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY)
 356         av_add_index_entry(st, pos, dts, size, 0, AVINDEX_KEYFRAME);
 357     break;
 358  }
 359
 360     // if not streamed and no duration from metadata then seek to end to find the duration from the timestamps
 361     if(!url_is_streamed(s->pb) && s->duration==AV_NOPTS_VALUE){
 362         int size;
 363         const int pos= url_ftell(s->pb);
 364         const int fsize= url_fsize(s->pb);
 365         url_fseek(s->pb, fsize-4, SEEK_SET);
 366         size= get_be32(s->pb);
 367         url_fseek(s->pb, fsize-3-size, SEEK_SET);
 368         if(size == get_be24(s->pb) + 11){
 369             s->duration= get_be24(s->pb) * (int64_t)AV_TIME_BASE / 1000;
 370         }
 371         url_fseek(s->pb, pos, SEEK_SET);
 372     }
 373
 374     if(is_audio){
 375         if(!st->codec->sample_rate || !st->codec->bits_per_sample || (!st->codec->codec_id && !st->codec->codec_tag)) {
 376             st->codec->channels = (flags & FLV_AUDIO_CHANNEL_MASK) == FLV_STEREO ? 2 : 1;
 377             if((flags & FLV_AUDIO_CODECID_MASK) == FLV_CODECID_NELLYMOSER_8HZ_MONO)
 378                 st->codec->sample_rate= 8000;
 379             else
 380                 st->codec->sample_rate = (44100 << ((flags & FLV_AUDIO_SAMPLERATE_MASK) >> FLV_AUDIO_SAMPLERATE_OFFSET) >> 3);
 381             st->codec->bits_per_sample = (flags & FLV_AUDIO_SAMPLESIZE_MASK) ? 16 : 8;
 382             flv_set_audio_codec(s, st, flags & FLV_AUDIO_CODECID_MASK);
 383         }
 384     }else{
 385         size -= flv_set_video_codec(s, st, flags & FLV_VIDEO_CODECID_MASK);
 386     }
 387
 388     if (st->codec->codec_id == CODEC_ID_AAC ||
 389         st->codec->codec_id == CODEC_ID_H264) {
 390         int type = get_byte(s->pb);
 391         size--;
 392         if (st->codec->codec_id == CODEC_ID_H264) {
 393             // cts offset ignored because it might to be signed
 394             // and would cause pts < dts
 395             get_be24(s->pb);
 396         }
 397         if (type == 0) {
 398             if ((ret = flv_get_extradata(s, st, size - 1)) < 0)
 399                 return ret;
 400             goto retry;
 401         }
 402     }
 403
 404     ret= av_get_packet(s->pb, pkt, size - 1);
 405     if (ret <= 0) {
 406         return AVERROR(EIO);
 407     }
 408     /* note: we need to modify the packet size here to handle the last
 409        packet */
 410     pkt->size = ret;
 411     pkt->dts = dts;
 412     pkt->stream_index = st->index;
 413
 414     if (is_audio || ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY))
 415         pkt->flags |= PKT_FLAG_KEY;
 416
 417     return ret;
 418 }
 419
 420 AVInputFormat flv_demuxer = {
 421     "flv",
 422     NULL_IF_CONFIG_SMALL("FLV format"),
 423     0,
 424     flv_probe,
 425     flv_read_header,
 426     flv_read_packet,
 427     .extensions = "flv",
 428     .value = CODEC_ID_FLV1,
 429 };