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 "libavutil/avstring.h"
  28 #include "libavcodec/bytestream.h"
  29 #include "libavcodec/mpeg4audio.h"
  30 #include "avformat.h"
  31 #include "flv.h"
  32
  33 typedef struct {
  34     int wrong_dts; ///< wrong dts due to negative cts
  35 } FLVContext;
  36
  37 static int flv_probe(AVProbeData *p)
  38 {
  39     const uint8_t *d;
  40
  41     d = p->buf;
  42     if (d[0] == 'F' && d[1] == 'L' && d[2] == 'V' && d[3] < 5 && d[5]==0 && AV_RB32(d+5)>8) {
  43         return AVPROBE_SCORE_MAX;
  44     }
  45     return 0;
  46 }
  47
  48 static void flv_set_audio_codec(AVFormatContext *s, AVStream *astream, int flv_codecid) {
  49     AVCodecContext *acodec = astream->codec;
  50     switch(flv_codecid) {
  51         //no distinction between S16 and S8 PCM codec flags
  52         case FLV_CODECID_PCM:
  53             acodec->codec_id = acodec->bits_per_coded_sample == 8 ? CODEC_ID_PCM_S8 :
  54 #if HAVE_BIGENDIAN
  55                                 CODEC_ID_PCM_S16BE;
  56 #else
  57                                 CODEC_ID_PCM_S16LE;
  58 #endif
  59             break;
  60         case FLV_CODECID_PCM_LE:
  61             acodec->codec_id = acodec->bits_per_coded_sample == 8 ? CODEC_ID_PCM_S8 : CODEC_ID_PCM_S16LE; break;
  62         case FLV_CODECID_AAC  : acodec->codec_id = CODEC_ID_AAC;                                    break;
  63         case FLV_CODECID_ADPCM: acodec->codec_id = CODEC_ID_ADPCM_SWF;                              break;
  64         case FLV_CODECID_SPEEX:
  65             acodec->codec_id = CODEC_ID_SPEEX;
  66             acodec->sample_rate = 16000;
  67             break;
  68         case FLV_CODECID_MP3  : acodec->codec_id = CODEC_ID_MP3      ; astream->need_parsing = AVSTREAM_PARSE_FULL; break;
  69         case FLV_CODECID_NELLYMOSER_8KHZ_MONO:
  70             acodec->sample_rate = 8000; //in case metadata does not otherwise declare samplerate
  71         case FLV_CODECID_NELLYMOSER:
  72             acodec->codec_id = CODEC_ID_NELLYMOSER;
  73             break;
  74         default:
  75             av_log(s, AV_LOG_INFO, "Unsupported audio codec (%x)\n", flv_codecid >> FLV_AUDIO_CODECID_OFFSET);
  76             acodec->codec_tag = flv_codecid >> FLV_AUDIO_CODECID_OFFSET;
  77     }
  78 }
  79
  80 static int flv_set_video_codec(AVFormatContext *s, AVStream *vstream, int flv_codecid) {
  81     AVCodecContext *vcodec = vstream->codec;
  82     switch(flv_codecid) {
  83         case FLV_CODECID_H263  : vcodec->codec_id = CODEC_ID_FLV1   ; break;
  84         case FLV_CODECID_SCREEN: vcodec->codec_id = CODEC_ID_FLASHSV; break;
  85         case FLV_CODECID_VP6   : vcodec->codec_id = CODEC_ID_VP6F   ;
  86         case FLV_CODECID_VP6A  :
  87             if(flv_codecid == FLV_CODECID_VP6A)
  88                 vcodec->codec_id = CODEC_ID_VP6A;
  89             if(vcodec->extradata_size != 1) {
  90                 vcodec->extradata_size = 1;
  91                 vcodec->extradata = av_malloc(1);
  92             }
  93             vcodec->extradata[0] = get_byte(s->pb);
  94             return 1; // 1 byte body size adjustment for flv_read_packet()
  95         case FLV_CODECID_H264:
  96             vcodec->codec_id = CODEC_ID_H264;
  97             return 3; // not 4, reading packet type will consume one byte
  98         default:
  99             av_log(s, AV_LOG_INFO, "Unsupported video codec (%x)\n", flv_codecid);
 100             vcodec->codec_tag = flv_codecid;
 101     }
 102
 103     return 0;
 104 }
 105
 106 static int amf_get_string(ByteIOContext *ioc, char *buffer, int buffsize) {
 107     int length = get_be16(ioc);
 108     if(length >= buffsize) {
 109         url_fskip(ioc, length);
 110         return -1;
 111     }
 112
 113     get_buffer(ioc, buffer, length);
 114
 115     buffer[length] = '\0';
 116
 117     return length;
 118 }
 119
 120 static int amf_parse_object(AVFormatContext *s, AVStream *astream, AVStream *vstream, const char *key, int64_t max_pos, int depth) {
 121     AVCodecContext *acodec, *vcodec;
 122     ByteIOContext *ioc;
 123     AMFDataType amf_type;
 124     char str_val[256];
 125     double num_val;
 126
 127     num_val = 0;
 128     ioc = s->pb;
 129
 130     amf_type = get_byte(ioc);
 131
 132     switch(amf_type) {
 133         case AMF_DATA_TYPE_NUMBER:
 134             num_val = av_int2dbl(get_be64(ioc)); break;
 135         case AMF_DATA_TYPE_BOOL:
 136             num_val = get_byte(ioc); break;
 137         case AMF_DATA_TYPE_STRING:
 138             if(amf_get_string(ioc, str_val, sizeof(str_val)) < 0)
 139                 return -1;
 140             break;
 141         case AMF_DATA_TYPE_OBJECT: {
 142             unsigned int keylen;
 143
 144             while(url_ftell(ioc) < max_pos - 2 && (keylen = get_be16(ioc))) {
 145                 url_fskip(ioc, keylen); //skip key string
 146                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 147                     return -1; //if we couldn't skip, bomb out.
 148             }
 149             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 150                 return -1;
 151         }
 152             break;
 153         case AMF_DATA_TYPE_NULL:
 154         case AMF_DATA_TYPE_UNDEFINED:
 155         case AMF_DATA_TYPE_UNSUPPORTED:
 156             break; //these take up no additional space
 157         case AMF_DATA_TYPE_MIXEDARRAY:
 158             url_fskip(ioc, 4); //skip 32-bit max array index
 159             while(url_ftell(ioc) < max_pos - 2 && amf_get_string(ioc, str_val, sizeof(str_val)) > 0) {
 160                 //this is the only case in which we would want a nested parse to not skip over the object
 161                 if(amf_parse_object(s, astream, vstream, str_val, max_pos, depth + 1) < 0)
 162                     return -1;
 163             }
 164             if(get_byte(ioc) != AMF_END_OF_OBJECT)
 165                 return -1;
 166             break;
 167         case AMF_DATA_TYPE_ARRAY: {
 168             unsigned int arraylen, i;
 169
 170             arraylen = get_be32(ioc);
 171             for(i = 0; i < arraylen && url_ftell(ioc) < max_pos - 1; i++) {
 172                 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
 173                     return -1; //if we couldn't skip, bomb out.
 174             }
 175         }
 176             break;
 177         case AMF_DATA_TYPE_DATE:
 178             url_fskip(ioc, 8 + 2); //timestamp (double) and UTC offset (int16)
 179             break;
 180         default: //unsupported type, we couldn't skip
 181             return -1;
 182     }
 183
 184     if(depth == 1 && key) { //only look for metadata values when we are not nested and key != NULL
 185         acodec = astream ? astream->codec : NULL;
 186         vcodec = vstream ? vstream->codec : NULL;
 187
 188         if(amf_type == AMF_DATA_TYPE_BOOL) {
 189             av_strlcpy(str_val, num_val > 0 ? "true" : "false", sizeof(str_val));
 190             av_metadata_set(&s->metadata, key, str_val);
 191         } else if(amf_type == AMF_DATA_TYPE_NUMBER) {
 192             snprintf(str_val, sizeof(str_val), "%.f", num_val);
 193             av_metadata_set(&s->metadata, key, str_val);
 194             if(!strcmp(key, "duration")) s->duration = num_val * AV_TIME_BASE;
 195             else if(!strcmp(key, "videodatarate") && vcodec && 0 <= (int)(num_val * 1024.0))
 196                 vcodec->bit_rate = num_val * 1024.0;
 197         } else if (amf_type == AMF_DATA_TYPE_STRING)
 198           av_metadata_set(&s->metadata, key, str_val);
 199     }
 200
 201     return 0;
 202 }
 203
 204 static int flv_read_metabody(AVFormatContext *s, int64_t next_pos) {
 205     AMFDataType type;
 206     AVStream *stream, *astream, *vstream;
 207     ByteIOContext *ioc;
 208     int i;
 209     char buffer[11]; //only needs to hold the string "onMetaData". Anything longer is something we don't want.
 210
 211     astream = NULL;
 212     vstream = NULL;
 213     ioc = s->pb;
 214
 215     //first object needs to be "onMetaData" string
 216     type = get_byte(ioc);
 217     if(type != AMF_DATA_TYPE_STRING || amf_get_string(ioc, buffer, sizeof(buffer)) < 0 || strcmp(buffer, "onMetaData"))
 218         return -1;
 219
 220     //find the streams now so that amf_parse_object doesn't need to do the lookup every time it is called.
 221     for(i = 0; i < s->nb_streams; i++) {
 222         stream = s->streams[i];
 223         if     (stream->codec->codec_type == CODEC_TYPE_AUDIO) astream = stream;
 224         else if(stream->codec->codec_type == CODEC_TYPE_VIDEO) vstream = stream;
 225     }
 226
 227     //parse the second object (we want a mixed array)
 228     if(amf_parse_object(s, astream, vstream, buffer, next_pos, 0) < 0)
 229         return -1;
 230
 231     return 0;
 232 }
 233
 234 static AVStream *create_stream(AVFormatContext *s, int is_audio){
 235     AVStream *st = av_new_stream(s, is_audio);
 236     if (!st)
 237         return NULL;
 238     st->codec->codec_type = is_audio ? CODEC_TYPE_AUDIO : CODEC_TYPE_VIDEO;
 239     av_set_pts_info(st, 32, 1, 1000); /* 32 bit pts in ms */
 240     return st;
 241 }
 242
 243 static int flv_read_header(AVFormatContext *s,
 244                            AVFormatParameters *ap)
 245 {
 246     int offset, flags;
 247
 248     url_fskip(s->pb, 4);
 249     flags = get_byte(s->pb);
 250     /* old flvtool cleared this field */
 251     /* FIXME: better fix needed */
 252     if (!flags) {
 253         flags = FLV_HEADER_FLAG_HASVIDEO | FLV_HEADER_FLAG_HASAUDIO;
 254         av_log(s, AV_LOG_WARNING, "Broken FLV file, which says no streams present, this might fail\n");
 255     }
 256
 257     if((flags & (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 258              != (FLV_HEADER_FLAG_HASVIDEO|FLV_HEADER_FLAG_HASAUDIO))
 259         s->ctx_flags |= AVFMTCTX_NOHEADER;
 260
 261     if(flags & FLV_HEADER_FLAG_HASVIDEO){
 262         if(!create_stream(s, 0))
 263             return AVERROR(ENOMEM);
 264     }
 265     if(flags & FLV_HEADER_FLAG_HASAUDIO){
 266         if(!create_stream(s, 1))
 267             return AVERROR(ENOMEM);
 268     }
 269
 270     offset = get_be32(s->pb);
 271     url_fseek(s->pb, offset, SEEK_SET);
 272
 273     s->start_time = 0;
 274
 275     return 0;
 276 }
 277
 278 static int flv_get_extradata(AVFormatContext *s, AVStream *st, int size)
 279 {
 280     av_free(st->codec->extradata);
 281     st->codec->extradata = av_mallocz(size + FF_INPUT_BUFFER_PADDING_SIZE);
 282     if (!st->codec->extradata)
 283         return AVERROR(ENOMEM);
 284     st->codec->extradata_size = size;
 285     get_buffer(s->pb, st->codec->extradata, st->codec->extradata_size);
 286     return 0;
 287 }
 288
 289 static int flv_read_packet(AVFormatContext *s, AVPacket *pkt)
 290 {
 291     FLVContext *flv = s->priv_data;
 292     int ret, i, type, size, flags, is_audio;
 293     int64_t next, pos;
 294     int64_t dts, pts = AV_NOPTS_VALUE;
 295     AVStream *st = NULL;
 296
 297  for(;;){
 298     pos = url_ftell(s->pb);
 299     url_fskip(s->pb, 4); /* size of previous packet */
 300     type = get_byte(s->pb);
 301     size = get_be24(s->pb);
 302     dts = get_be24(s->pb);
 303     dts |= get_byte(s->pb) << 24;
 304 //    av_log(s, AV_LOG_DEBUG, "type:%d, size:%d, dts:%d\n", type, size, dts);
 305     if (url_feof(s->pb))
 306         return AVERROR_EOF;
 307     url_fskip(s->pb, 3); /* stream id, always 0 */
 308     flags = 0;
 309
 310     if(size == 0)
 311         continue;
 312
 313     next= size + url_ftell(s->pb);
 314
 315     if (type == FLV_TAG_TYPE_AUDIO) {
 316         is_audio=1;
 317         flags = get_byte(s->pb);
 318         size--;
 319     } else if (type == FLV_TAG_TYPE_VIDEO) {
 320         is_audio=0;
 321         flags = get_byte(s->pb);
 322         size--;
 323         if ((flags & 0xf0) == 0x50) /* video info / command frame */
 324             goto skip;
 325     } else {
 326         if (type == FLV_TAG_TYPE_META && size > 13+1+4)
 327             flv_read_metabody(s, next);
 328         else /* skip packet */
 329             av_log(s, AV_LOG_DEBUG, "skipping flv packet: type %d, size %d, flags %d\n", type, size, flags);
 330     skip:
 331         url_fseek(s->pb, next, SEEK_SET);
 332         continue;
 333     }
 334
 335     /* skip empty data packets */
 336     if (!size)
 337         continue;
 338
 339     /* now find stream */
 340     for(i=0;i<s->nb_streams;i++) {
 341         st = s->streams[i];
 342         if (st->id == is_audio)
 343             break;
 344     }
 345     if(i == s->nb_streams){
 346         av_log(s, AV_LOG_ERROR, "invalid stream\n");
 347         st= create_stream(s, is_audio);
 348         s->ctx_flags &= ~AVFMTCTX_NOHEADER;
 349     }
 350 //    av_log(s, AV_LOG_DEBUG, "%d %X %d \n", is_audio, flags, st->discard);
 351     if(  (st->discard >= AVDISCARD_NONKEY && !((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY ||         is_audio))
 352        ||(st->discard >= AVDISCARD_BIDIR  &&  ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_DISP_INTER && !is_audio))
 353        || st->discard >= AVDISCARD_ALL
 354        ){
 355         url_fseek(s->pb, next, SEEK_SET);
 356         continue;
 357     }
 358     if ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY)
 359         av_add_index_entry(st, pos, dts, size, 0, AVINDEX_KEYFRAME);
 360     break;
 361  }
 362
 363     // if not streamed and no duration from metadata then seek to end to find the duration from the timestamps
 364     if(!url_is_streamed(s->pb) && s->duration==AV_NOPTS_VALUE){
 365         int size;
 366         const int64_t pos= url_ftell(s->pb);
 367         const int64_t fsize= url_fsize(s->pb);
 368         url_fseek(s->pb, fsize-4, SEEK_SET);
 369         size= get_be32(s->pb);
 370         url_fseek(s->pb, fsize-3-size, SEEK_SET);
 371         if(size == get_be24(s->pb) + 11){
 372             uint32_t ts = get_be24(s->pb);
 373             ts |= get_byte(s->pb) << 24;
 374             s->duration = ts * (int64_t)AV_TIME_BASE / 1000;
 375         }
 376         url_fseek(s->pb, pos, SEEK_SET);
 377     }
 378
 379     if(is_audio){
 380         if(!st->codec->channels || !st->codec->sample_rate || !st->codec->bits_per_coded_sample) {
 381             st->codec->channels = (flags & FLV_AUDIO_CHANNEL_MASK) == FLV_STEREO ? 2 : 1;
 382             st->codec->sample_rate = (44100 << ((flags & FLV_AUDIO_SAMPLERATE_MASK) >> FLV_AUDIO_SAMPLERATE_OFFSET) >> 3);
 383             st->codec->bits_per_coded_sample = (flags & FLV_AUDIO_SAMPLESIZE_MASK) ? 16 : 8;
 384         }
 385         if(!st->codec->codec_id){
 386             flv_set_audio_codec(s, st, flags & FLV_AUDIO_CODECID_MASK);
 387         }
 388     }else{
 389         size -= flv_set_video_codec(s, st, flags & FLV_VIDEO_CODECID_MASK);
 390     }
 391
 392     if (st->codec->codec_id == CODEC_ID_AAC ||
 393         st->codec->codec_id == CODEC_ID_H264) {
 394         int type = get_byte(s->pb);
 395         size--;
 396         if (st->codec->codec_id == CODEC_ID_H264) {
 397             int32_t cts = (get_be24(s->pb)+0xff800000)^0xff800000; // sign extension
 398             pts = dts + cts;
 399             if (cts < 0) { // dts are wrong
 400                 flv->wrong_dts = 1;
 401                 av_log(s, AV_LOG_WARNING, "negative cts, previous timestamps might be wrong\n");
 402             }
 403             if (flv->wrong_dts)
 404                 dts = AV_NOPTS_VALUE;
 405         }
 406         if (type == 0) {
 407             if ((ret = flv_get_extradata(s, st, size)) < 0)
 408                 return ret;
 409             if (st->codec->codec_id == CODEC_ID_AAC) {
 410                 MPEG4AudioConfig cfg;
 411                 ff_mpeg4audio_get_config(&cfg, st->codec->extradata,
 412                                          st->codec->extradata_size);
 413                 if (cfg.chan_config > 7)
 414                     return -1;
 415                 st->codec->channels = ff_mpeg4audio_channels[cfg.chan_config];
 416                 st->codec->sample_rate = cfg.sample_rate;
 417                 dprintf(s, "mp4a config channels %d sample rate %d\n",
 418                         st->codec->channels, st->codec->sample_rate);
 419             }
 420
 421             return AVERROR(EAGAIN);
 422         }
 423     }
 424
 425     /* skip empty data packets */
 426     if (!size)
 427         return AVERROR(EAGAIN);
 428
 429     ret= av_get_packet(s->pb, pkt, size);
 430     if (ret < 0) {
 431         return AVERROR(EIO);
 432     }
 433     /* note: we need to modify the packet size here to handle the last
 434        packet */
 435     pkt->size = ret;
 436     pkt->dts = dts;
 437     pkt->pts = pts == AV_NOPTS_VALUE ? dts : pts;
 438     pkt->stream_index = st->index;
 439
 440     if (is_audio || ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY))
 441         pkt->flags |= PKT_FLAG_KEY;
 442
 443     return ret;
 444 }
 445
 446 AVInputFormat flv_demuxer = {
 447     "flv",
 448     NULL_IF_CONFIG_SMALL("FLV format"),
 449     sizeof(FLVContext),
 450     flv_probe,
 451     flv_read_header,
 452     flv_read_packet,
 453     .extensions = "flv",
 454     .value = CODEC_ID_FLV1,
 455 };