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