libavformat/rtpdec.c

   1 /*
   2  * RTP input format
   3  * Copyright (c) 2002 Fabrice Bellard.
   4  *
   5  * This file is part of FFmpeg.
   6  *
   7  * FFmpeg is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * FFmpeg is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with FFmpeg; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 #include "libavcodec/bitstream.h"
  23 #include "avformat.h"
  24 #include "mpegts.h"
  25
  26 #include <unistd.h>
  27 #include "network.h"
  28
  29 #include "rtp_internal.h"
  30 #include "rtp_h264.h"
  31
  32 //#define DEBUG
  33
  34 /* TODO: - add RTCP statistics reporting (should be optional).
  35
  36          - add support for h263/mpeg4 packetized output : IDEA: send a
  37          buffer to 'rtp_write_packet' contains all the packets for ONE
  38          frame. Each packet should have a four byte header containing
  39          the length in big endian format (same trick as
  40          'url_open_dyn_packet_buf')
  41 */
  42
  43 /* statistics functions */
  44 RTPDynamicProtocolHandler *RTPFirstDynamicPayloadHandler= NULL;
  45
  46 static RTPDynamicProtocolHandler mp4v_es_handler= {"MP4V-ES", CODEC_TYPE_VIDEO, CODEC_ID_MPEG4};
  47 static RTPDynamicProtocolHandler mpeg4_generic_handler= {"mpeg4-generic", CODEC_TYPE_AUDIO, CODEC_ID_AAC};
  48
  49 static void register_dynamic_payload_handler(RTPDynamicProtocolHandler *handler)
  50 {
  51     handler->next= RTPFirstDynamicPayloadHandler;
  52     RTPFirstDynamicPayloadHandler= handler;
  53 }
  54
  55 void av_register_rtp_dynamic_payload_handlers(void)
  56 {
  57     register_dynamic_payload_handler(&mp4v_es_handler);
  58     register_dynamic_payload_handler(&mpeg4_generic_handler);
  59     register_dynamic_payload_handler(&ff_h264_dynamic_handler);
  60 }
  61
  62 static int rtcp_parse_packet(RTPDemuxContext *s, const unsigned char *buf, int len)
  63 {
  64     if (buf[1] != 200)
  65         return -1;
  66     s->last_rtcp_ntp_time = AV_RB64(buf + 8);
  67     if (s->first_rtcp_ntp_time == AV_NOPTS_VALUE)
  68         s->first_rtcp_ntp_time = s->last_rtcp_ntp_time;
  69     s->last_rtcp_timestamp = AV_RB32(buf + 16);
  70     return 0;
  71 }
  72
  73 #define RTP_SEQ_MOD (1<<16)
  74
  75 /**
  76 * called on parse open packet
  77 */
  78 static void rtp_init_statistics(RTPStatistics *s, uint16_t base_sequence) // called on parse open packet.
  79 {
  80     memset(s, 0, sizeof(RTPStatistics));
  81     s->max_seq= base_sequence;
  82     s->probation= 1;
  83 }
  84
  85 /**
  86 * called whenever there is a large jump in sequence numbers, or when they get out of probation...
  87 */
  88 static void rtp_init_sequence(RTPStatistics *s, uint16_t seq)
  89 {
  90     s->max_seq= seq;
  91     s->cycles= 0;
  92     s->base_seq= seq -1;
  93     s->bad_seq= RTP_SEQ_MOD + 1;
  94     s->received= 0;
  95     s->expected_prior= 0;
  96     s->received_prior= 0;
  97     s->jitter= 0;
  98     s->transit= 0;
  99 }
 100
 101 /**
 102 * returns 1 if we should handle this packet.
 103 */
 104 static int rtp_valid_packet_in_sequence(RTPStatistics *s, uint16_t seq)
 105 {
 106     uint16_t udelta= seq - s->max_seq;
 107     const int MAX_DROPOUT= 3000;
 108     const int MAX_MISORDER = 100;
 109     const int MIN_SEQUENTIAL = 2;
 110
 111     /* source not valid until MIN_SEQUENTIAL packets with sequence seq. numbers have been received */
 112     if(s->probation)
 113     {
 114         if(seq==s->max_seq + 1) {
 115             s->probation--;
 116             s->max_seq= seq;
 117             if(s->probation==0) {
 118                 rtp_init_sequence(s, seq);
 119                 s->received++;
 120                 return 1;
 121             }
 122         } else {
 123             s->probation= MIN_SEQUENTIAL - 1;
 124             s->max_seq = seq;
 125         }
 126     } else if (udelta < MAX_DROPOUT) {
 127         // in order, with permissible gap
 128         if(seq < s->max_seq) {
 129             //sequence number wrapped; count antother 64k cycles
 130             s->cycles += RTP_SEQ_MOD;
 131         }
 132         s->max_seq= seq;
 133     } else if (udelta <= RTP_SEQ_MOD - MAX_MISORDER) {
 134         // sequence made a large jump...
 135         if(seq==s->bad_seq) {
 136             // two sequential packets-- assume that the other side restarted without telling us; just resync.
 137             rtp_init_sequence(s, seq);
 138         } else {
 139             s->bad_seq= (seq + 1) & (RTP_SEQ_MOD-1);
 140             return 0;
 141         }
 142     } else {
 143         // duplicate or reordered packet...
 144     }
 145     s->received++;
 146     return 1;
 147 }
 148
 149 #if 0
 150 /**
 151 * This function is currently unused; without a valid local ntp time, I don't see how we could calculate the
 152 * difference between the arrival and sent timestamp.  As a result, the jitter and transit statistics values
 153 * never change.  I left this in in case someone else can see a way. (rdm)
 154 */
 155 static void rtcp_update_jitter(RTPStatistics *s, uint32_t sent_timestamp, uint32_t arrival_timestamp)
 156 {
 157     uint32_t transit= arrival_timestamp - sent_timestamp;
 158     int d;
 159     s->transit= transit;
 160     d= FFABS(transit - s->transit);
 161     s->jitter += d - ((s->jitter + 8)>>4);
 162 }
 163 #endif
 164
 165 int rtp_check_and_send_back_rr(RTPDemuxContext *s, int count)
 166 {
 167     ByteIOContext *pb;
 168     uint8_t *buf;
 169     int len;
 170     int rtcp_bytes;
 171     RTPStatistics *stats= &s->statistics;
 172     uint32_t lost;
 173     uint32_t extended_max;
 174     uint32_t expected_interval;
 175     uint32_t received_interval;
 176     uint32_t lost_interval;
 177     uint32_t expected;
 178     uint32_t fraction;
 179     uint64_t ntp_time= s->last_rtcp_ntp_time; // TODO: Get local ntp time?
 180
 181     if (!s->rtp_ctx || (count < 1))
 182         return -1;
 183
 184     /* TODO: I think this is way too often; RFC 1889 has algorithm for this */
 185     /* XXX: mpeg pts hardcoded. RTCP send every 0.5 seconds */
 186     s->octet_count += count;
 187     rtcp_bytes = ((s->octet_count - s->last_octet_count) * RTCP_TX_RATIO_NUM) /
 188         RTCP_TX_RATIO_DEN;
 189     rtcp_bytes /= 50; // mmu_man: that's enough for me... VLC sends much less btw !?
 190     if (rtcp_bytes < 28)
 191         return -1;
 192     s->last_octet_count = s->octet_count;
 193
 194     if (url_open_dyn_buf(&pb) < 0)
 195         return -1;
 196
 197     // Receiver Report
 198     put_byte(pb, (RTP_VERSION << 6) + 1); /* 1 report block */
 199     put_byte(pb, 201);
 200     put_be16(pb, 7); /* length in words - 1 */
 201     put_be32(pb, s->ssrc); // our own SSRC
 202     put_be32(pb, s->ssrc); // XXX: should be the server's here!
 203     // some placeholders we should really fill...
 204     // RFC 1889/p64
 205     extended_max= stats->cycles + stats->max_seq;
 206     expected= extended_max - stats->base_seq + 1;
 207     lost= expected - stats->received;
 208     lost= FFMIN(lost, 0xffffff); // clamp it since it's only 24 bits...
 209     expected_interval= expected - stats->expected_prior;
 210     stats->expected_prior= expected;
 211     received_interval= stats->received - stats->received_prior;
 212     stats->received_prior= stats->received;
 213     lost_interval= expected_interval - received_interval;
 214     if (expected_interval==0 || lost_interval<=0) fraction= 0;
 215     else fraction = (lost_interval<<8)/expected_interval;
 216
 217     fraction= (fraction<<24) | lost;
 218
 219     put_be32(pb, fraction); /* 8 bits of fraction, 24 bits of total packets lost */
 220     put_be32(pb, extended_max); /* max sequence received */
 221     put_be32(pb, stats->jitter>>4); /* jitter */
 222
 223     if(s->last_rtcp_ntp_time==AV_NOPTS_VALUE)
 224     {
 225         put_be32(pb, 0); /* last SR timestamp */
 226         put_be32(pb, 0); /* delay since last SR */
 227     } else {
 228         uint32_t middle_32_bits= s->last_rtcp_ntp_time>>16; // this is valid, right? do we need to handle 64 bit values special?
 229         uint32_t delay_since_last= ntp_time - s->last_rtcp_ntp_time;
 230
 231         put_be32(pb, middle_32_bits); /* last SR timestamp */
 232         put_be32(pb, delay_since_last); /* delay since last SR */
 233     }
 234
 235     // CNAME
 236     put_byte(pb, (RTP_VERSION << 6) + 1); /* 1 report block */
 237     put_byte(pb, 202);
 238     len = strlen(s->hostname);
 239     put_be16(pb, (6 + len + 3) / 4); /* length in words - 1 */
 240     put_be32(pb, s->ssrc);
 241     put_byte(pb, 0x01);
 242     put_byte(pb, len);
 243     put_buffer(pb, s->hostname, len);
 244     // padding
 245     for (len = (6 + len) % 4; len % 4; len++) {
 246         put_byte(pb, 0);
 247     }
 248
 249     put_flush_packet(pb);
 250     len = url_close_dyn_buf(pb, &buf);
 251     if ((len > 0) && buf) {
 252         int result;
 253         dprintf(s->ic, "sending %d bytes of RR\n", len);
 254         result= url_write(s->rtp_ctx, buf, len);
 255         dprintf(s->ic, "result from url_write: %d\n", result);
 256         av_free(buf);
 257     }
 258     return 0;
 259 }
 260
 261 /**
 262  * open a new RTP parse context for stream 'st'. 'st' can be NULL for
 263  * MPEG2TS streams to indicate that they should be demuxed inside the
 264  * rtp demux (otherwise CODEC_ID_MPEG2TS packets are returned)
 265  * TODO: change this to not take rtp_payload data, and use the new dynamic payload system.
 266  */
 267 RTPDemuxContext *rtp_parse_open(AVFormatContext *s1, AVStream *st, URLContext *rtpc, int payload_type, rtp_payload_data_t *rtp_payload_data)
 268 {
 269     RTPDemuxContext *s;
 270
 271     s = av_mallocz(sizeof(RTPDemuxContext));
 272     if (!s)
 273         return NULL;
 274     s->payload_type = payload_type;
 275     s->last_rtcp_ntp_time = AV_NOPTS_VALUE;
 276     s->first_rtcp_ntp_time = AV_NOPTS_VALUE;
 277     s->ic = s1;
 278     s->st = st;
 279     s->rtp_payload_data = rtp_payload_data;
 280     rtp_init_statistics(&s->statistics, 0); // do we know the initial sequence from sdp?
 281     av_set_pts_info(s->st, 32, 1, 90000);
 282     if (!strcmp(ff_rtp_enc_name(payload_type), "MP2T")) {
 283         s->ts = mpegts_parse_open(s->ic);
 284         if (s->ts == NULL) {
 285             av_free(s);
 286             return NULL;
 287         }
 288     } else {
 289         switch(st->codec->codec_id) {
 290         case CODEC_ID_MPEG1VIDEO:
 291         case CODEC_ID_MPEG2VIDEO:
 292         case CODEC_ID_MP2:
 293         case CODEC_ID_MP3:
 294         case CODEC_ID_MPEG4:
 295         case CODEC_ID_H264:
 296             st->need_parsing = AVSTREAM_PARSE_FULL;
 297             break;
 298         default:
 299             if (st->codec->codec_type == CODEC_TYPE_AUDIO) {
 300                 av_set_pts_info(st, 32, 1, st->codec->sample_rate);
 301             }
 302             break;
 303         }
 304     }
 305     // needed to send back RTCP RR in RTSP sessions
 306     s->rtp_ctx = rtpc;
 307     gethostname(s->hostname, sizeof(s->hostname));
 308     return s;
 309 }
 310
 311 static int rtp_parse_mp4_au(RTPDemuxContext *s, const uint8_t *buf)
 312 {
 313     int au_headers_length, au_header_size, i;
 314     GetBitContext getbitcontext;
 315     rtp_payload_data_t *infos;
 316
 317     infos = s->rtp_payload_data;
 318
 319     if (infos == NULL)
 320         return -1;
 321
 322     /* decode the first 2 bytes where the AUHeader sections are stored
 323        length in bits */
 324     au_headers_length = AV_RB16(buf);
 325
 326     if (au_headers_length > RTP_MAX_PACKET_LENGTH)
 327       return -1;
 328
 329     infos->au_headers_length_bytes = (au_headers_length + 7) / 8;
 330
 331     /* skip AU headers length section (2 bytes) */
 332     buf += 2;
 333
 334     init_get_bits(&getbitcontext, buf, infos->au_headers_length_bytes * 8);
 335
 336     /* XXX: Wrong if optionnal additional sections are present (cts, dts etc...) */
 337     au_header_size = infos->sizelength + infos->indexlength;
 338     if (au_header_size <= 0 || (au_headers_length % au_header_size != 0))
 339         return -1;
 340
 341     infos->nb_au_headers = au_headers_length / au_header_size;
 342     infos->au_headers = av_malloc(sizeof(struct AUHeaders) * infos->nb_au_headers);
 343
 344     /* XXX: We handle multiple AU Section as only one (need to fix this for interleaving)
 345        In my test, the FAAD decoder does not behave correctly when sending each AU one by one
 346        but does when sending the whole as one big packet...  */
 347     infos->au_headers[0].size = 0;
 348     infos->au_headers[0].index = 0;
 349     for (i = 0; i < infos->nb_au_headers; ++i) {
 350         infos->au_headers[0].size += get_bits_long(&getbitcontext, infos->sizelength);
 351         infos->au_headers[0].index = get_bits_long(&getbitcontext, infos->indexlength);
 352     }
 353
 354     infos->nb_au_headers = 1;
 355
 356     return 0;
 357 }
 358
 359 /**
 360  * This was the second switch in rtp_parse packet.  Normalizes time, if required, sets stream_index, etc.
 361  */
 362 static void finalize_packet(RTPDemuxContext *s, AVPacket *pkt, uint32_t timestamp)
 363 {
 364     if (s->last_rtcp_ntp_time != AV_NOPTS_VALUE) {
 365         int64_t addend;
 366         int delta_timestamp;
 367
 368         /* compute pts from timestamp with received ntp_time */
 369         delta_timestamp = timestamp - s->last_rtcp_timestamp;
 370         /* convert to the PTS timebase */
 371         addend = av_rescale(s->last_rtcp_ntp_time - s->first_rtcp_ntp_time, s->st->time_base.den, (uint64_t)s->st->time_base.num << 32);
 372         pkt->pts = addend + delta_timestamp;
 373     }
 374     pkt->stream_index = s->st->index;
 375 }
 376
 377 /**
 378  * Parse an RTP or RTCP packet directly sent as a buffer.
 379  * @param s RTP parse context.
 380  * @param pkt returned packet
 381  * @param buf input buffer or NULL to read the next packets
 382  * @param len buffer len
 383  * @return 0 if a packet is returned, 1 if a packet is returned and more can follow
 384  * (use buf as NULL to read the next). -1 if no packet (error or no more packet).
 385  */
 386 int rtp_parse_packet(RTPDemuxContext *s, AVPacket *pkt,
 387                      const uint8_t *buf, int len)
 388 {
 389     unsigned int ssrc, h;
 390     int payload_type, seq, ret, flags = 0;
 391     AVStream *st;
 392     uint32_t timestamp;
 393     int rv= 0;
 394
 395     if (!buf) {
 396         /* return the next packets, if any */
 397         if(s->st && s->parse_packet) {
 398             timestamp= 0; ///< Should not be used if buf is NULL, but should be set to the timestamp of the packet returned....
 399             rv= s->parse_packet(s, pkt, &timestamp, NULL, 0, flags);
 400             finalize_packet(s, pkt, timestamp);
 401             return rv;
 402         } else {
 403             // TODO: Move to a dynamic packet handler (like above)
 404             if (s->read_buf_index >= s->read_buf_size)
 405                 return -1;
 406             ret = mpegts_parse_packet(s->ts, pkt, s->buf + s->read_buf_index,
 407                                       s->read_buf_size - s->read_buf_index);
 408             if (ret < 0)
 409                 return -1;
 410             s->read_buf_index += ret;
 411             if (s->read_buf_index < s->read_buf_size)
 412                 return 1;
 413             else
 414                 return 0;
 415         }
 416     }
 417
 418     if (len < 12)
 419         return -1;
 420
 421     if ((buf[0] & 0xc0) != (RTP_VERSION << 6))
 422         return -1;
 423     if (buf[1] >= 200 && buf[1] <= 204) {
 424         rtcp_parse_packet(s, buf, len);
 425         return -1;
 426     }
 427     payload_type = buf[1] & 0x7f;
 428     seq  = AV_RB16(buf + 2);
 429     timestamp = AV_RB32(buf + 4);
 430     ssrc = AV_RB32(buf + 8);
 431     /* store the ssrc in the RTPDemuxContext */
 432     s->ssrc = ssrc;
 433
 434     /* NOTE: we can handle only one payload type */
 435     if (s->payload_type != payload_type)
 436         return -1;
 437
 438     st = s->st;
 439     // only do something with this if all the rtp checks pass...
 440     if(!rtp_valid_packet_in_sequence(&s->statistics, seq))
 441     {
 442         av_log(st?st->codec:NULL, AV_LOG_ERROR, "RTP: PT=%02x: bad cseq %04x expected=%04x\n",
 443                payload_type, seq, ((s->seq + 1) & 0xffff));
 444         return -1;
 445     }
 446
 447     s->seq = seq;
 448     len -= 12;
 449     buf += 12;
 450
 451     if (!st) {
 452         /* specific MPEG2TS demux support */
 453         ret = mpegts_parse_packet(s->ts, pkt, buf, len);
 454         if (ret < 0)
 455             return -1;
 456         if (ret < len) {
 457             s->read_buf_size = len - ret;
 458             memcpy(s->buf, buf + ret, s->read_buf_size);
 459             s->read_buf_index = 0;
 460             return 1;
 461         }
 462     } else if (s->parse_packet) {
 463         rv = s->parse_packet(s, pkt, &timestamp, buf, len, flags);
 464     } else {
 465         // at this point, the RTP header has been stripped;  This is ASSUMING that there is only 1 CSRC, which in't wise.
 466         switch(st->codec->codec_id) {
 467         case CODEC_ID_MP2:
 468             /* better than nothing: skip mpeg audio RTP header */
 469             if (len <= 4)
 470                 return -1;
 471             h = AV_RB32(buf);
 472             len -= 4;
 473             buf += 4;
 474             av_new_packet(pkt, len);
 475             memcpy(pkt->data, buf, len);
 476             break;
 477         case CODEC_ID_MPEG1VIDEO:
 478         case CODEC_ID_MPEG2VIDEO:
 479             /* better than nothing: skip mpeg video RTP header */
 480             if (len <= 4)
 481                 return -1;
 482             h = AV_RB32(buf);
 483             buf += 4;
 484             len -= 4;
 485             if (h & (1 << 26)) {
 486                 /* mpeg2 */
 487                 if (len <= 4)
 488                     return -1;
 489                 buf += 4;
 490                 len -= 4;
 491             }
 492             av_new_packet(pkt, len);
 493             memcpy(pkt->data, buf, len);
 494             break;
 495             // moved from below, verbatim.  this is because this section handles packets, and the lower switch handles
 496             // timestamps.
 497             // TODO: Put this into a dynamic packet handler...
 498         case CODEC_ID_AAC:
 499             if (rtp_parse_mp4_au(s, buf))
 500                 return -1;
 501             {
 502                 rtp_payload_data_t *infos = s->rtp_payload_data;
 503                 if (infos == NULL)
 504                     return -1;
 505                 buf += infos->au_headers_length_bytes + 2;
 506                 len -= infos->au_headers_length_bytes + 2;
 507
 508                 /* XXX: Fixme we only handle the case where rtp_parse_mp4_au define
 509                     one au_header */
 510                 av_new_packet(pkt, infos->au_headers[0].size);
 511                 memcpy(pkt->data, buf, infos->au_headers[0].size);
 512                 buf += infos->au_headers[0].size;
 513                 len -= infos->au_headers[0].size;
 514             }
 515             s->read_buf_size = len;
 516             rv= 0;
 517             break;
 518         default:
 519             av_new_packet(pkt, len);
 520             memcpy(pkt->data, buf, len);
 521             break;
 522         }
 523
 524         // now perform timestamp things....
 525         finalize_packet(s, pkt, timestamp);
 526     }
 527     return rv;
 528 }
 529
 530 void rtp_parse_close(RTPDemuxContext *s)
 531 {
 532     // TODO: fold this into the protocol specific data fields.
 533     if (!strcmp(ff_rtp_enc_name(s->payload_type), "MP2T")) {
 534         mpegts_parse_close(s->ts);
 535     }
 536     av_free(s);
 537 }