Fix A/V synch for RTP streams that do not contain MPEG1 or 2
[libav.git] / libavformat / rtpdec.c
CommitLineData
8eb793c4
LA
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 */
245976da
DB
21
22#include "libavcodec/bitstream.h"
8eb793c4
LA
23#include "avformat.h"
24#include "mpegts.h"
8eb793c4
LA
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 */
44RTPDynamicProtocolHandler *RTPFirstDynamicPayloadHandler= NULL;
45
46static RTPDynamicProtocolHandler mp4v_es_handler= {"MP4V-ES", CODEC_TYPE_VIDEO, CODEC_ID_MPEG4};
47static RTPDynamicProtocolHandler mpeg4_generic_handler= {"mpeg4-generic", CODEC_TYPE_AUDIO, CODEC_ID_AAC};
48
49static void register_dynamic_payload_handler(RTPDynamicProtocolHandler *handler)
50{
51 handler->next= RTPFirstDynamicPayloadHandler;
52 RTPFirstDynamicPayloadHandler= handler;
53}
54
55void 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
62static 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*/
78static 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*/
88static 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*/
104static 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*/
155static 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
165int 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#if defined(DEBUG)
254 printf("sending %d bytes of RR\n", len);
255#endif
256 result= url_write(s->rtp_ctx, buf, len);
257#if defined(DEBUG)
258 printf("result from url_write: %d\n", result);
259#endif
260 av_free(buf);
261 }
262 return 0;
263}
264
265/**
266 * open a new RTP parse context for stream 'st'. 'st' can be NULL for
267 * MPEG2TS streams to indicate that they should be demuxed inside the
268 * rtp demux (otherwise CODEC_ID_MPEG2TS packets are returned)
269 * TODO: change this to not take rtp_payload data, and use the new dynamic payload system.
270 */
271RTPDemuxContext *rtp_parse_open(AVFormatContext *s1, AVStream *st, URLContext *rtpc, int payload_type, rtp_payload_data_t *rtp_payload_data)
272{
273 RTPDemuxContext *s;
274
275 s = av_mallocz(sizeof(RTPDemuxContext));
276 if (!s)
277 return NULL;
278 s->payload_type = payload_type;
279 s->last_rtcp_ntp_time = AV_NOPTS_VALUE;
280 s->first_rtcp_ntp_time = AV_NOPTS_VALUE;
281 s->ic = s1;
282 s->st = st;
283 s->rtp_payload_data = rtp_payload_data;
284 rtp_init_statistics(&s->statistics, 0); // do we know the initial sequence from sdp?
d6b9e57a 285 av_set_pts_info(s->st, 32, 1, 90000);
8eb793c4
LA
286 if (!strcmp(ff_rtp_enc_name(payload_type), "MP2T")) {
287 s->ts = mpegts_parse_open(s->ic);
288 if (s->ts == NULL) {
289 av_free(s);
290 return NULL;
291 }
292 } else {
293 switch(st->codec->codec_id) {
294 case CODEC_ID_MPEG1VIDEO:
295 case CODEC_ID_MPEG2VIDEO:
296 case CODEC_ID_MP2:
297 case CODEC_ID_MP3:
298 case CODEC_ID_MPEG4:
299 case CODEC_ID_H264:
300 st->need_parsing = AVSTREAM_PARSE_FULL;
301 break;
302 default:
d6b9e57a
LA
303 if (st->codec->codec_type == CODEC_TYPE_AUDIO) {
304 av_set_pts_info(st, 32, 1, st->codec->sample_rate);
305 }
8eb793c4
LA
306 break;
307 }
308 }
309 // needed to send back RTCP RR in RTSP sessions
310 s->rtp_ctx = rtpc;
311 gethostname(s->hostname, sizeof(s->hostname));
312 return s;
313}
314
315static int rtp_parse_mp4_au(RTPDemuxContext *s, const uint8_t *buf)
316{
317 int au_headers_length, au_header_size, i;
318 GetBitContext getbitcontext;
319 rtp_payload_data_t *infos;
320
321 infos = s->rtp_payload_data;
322
323 if (infos == NULL)
324 return -1;
325
bd107136 326 /* decode the first 2 bytes where the AUHeader sections are stored
8eb793c4
LA
327 length in bits */
328 au_headers_length = AV_RB16(buf);
329
330 if (au_headers_length > RTP_MAX_PACKET_LENGTH)
331 return -1;
332
333 infos->au_headers_length_bytes = (au_headers_length + 7) / 8;
334
335 /* skip AU headers length section (2 bytes) */
336 buf += 2;
337
338 init_get_bits(&getbitcontext, buf, infos->au_headers_length_bytes * 8);
339
340 /* XXX: Wrong if optionnal additional sections are present (cts, dts etc...) */
341 au_header_size = infos->sizelength + infos->indexlength;
342 if (au_header_size <= 0 || (au_headers_length % au_header_size != 0))
343 return -1;
344
345 infos->nb_au_headers = au_headers_length / au_header_size;
346 infos->au_headers = av_malloc(sizeof(struct AUHeaders) * infos->nb_au_headers);
347
348 /* XXX: We handle multiple AU Section as only one (need to fix this for interleaving)
349 In my test, the FAAD decoder does not behave correctly when sending each AU one by one
350 but does when sending the whole as one big packet... */
351 infos->au_headers[0].size = 0;
352 infos->au_headers[0].index = 0;
353 for (i = 0; i < infos->nb_au_headers; ++i) {
354 infos->au_headers[0].size += get_bits_long(&getbitcontext, infos->sizelength);
355 infos->au_headers[0].index = get_bits_long(&getbitcontext, infos->indexlength);
356 }
357
358 infos->nb_au_headers = 1;
359
360 return 0;
361}
362
363/**
364 * This was the second switch in rtp_parse packet. Normalizes time, if required, sets stream_index, etc.
365 */
366static void finalize_packet(RTPDemuxContext *s, AVPacket *pkt, uint32_t timestamp)
367{
8eb793c4
LA
368 if (s->last_rtcp_ntp_time != AV_NOPTS_VALUE) {
369 int64_t addend;
370
371 int delta_timestamp;
8eb793c4
LA
372 /* compute pts from timestamp with received ntp_time */
373 delta_timestamp = timestamp - s->last_rtcp_timestamp;
d6b9e57a
LA
374 /* convert to the PTS timebase */
375 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);
8eb793c4
LA
376 pkt->pts = addend + delta_timestamp;
377 }
8eb793c4
LA
378 pkt->stream_index = s->st->index;
379}
380
381/**
382 * Parse an RTP or RTCP packet directly sent as a buffer.
383 * @param s RTP parse context.
384 * @param pkt returned packet
385 * @param buf input buffer or NULL to read the next packets
386 * @param len buffer len
387 * @return 0 if a packet is returned, 1 if a packet is returned and more can follow
388 * (use buf as NULL to read the next). -1 if no packet (error or no more packet).
389 */
390int rtp_parse_packet(RTPDemuxContext *s, AVPacket *pkt,
391 const uint8_t *buf, int len)
392{
393 unsigned int ssrc, h;
f841a0fc 394 int payload_type, seq, ret, flags = 0;
8eb793c4
LA
395 AVStream *st;
396 uint32_t timestamp;
397 int rv= 0;
398
399 if (!buf) {
400 /* return the next packets, if any */
401 if(s->st && s->parse_packet) {
402 timestamp= 0; ///< Should not be used if buf is NULL, but should be set to the timestamp of the packet returned....
f841a0fc 403 rv= s->parse_packet(s, pkt, &timestamp, NULL, 0, flags);
8eb793c4
LA
404 finalize_packet(s, pkt, timestamp);
405 return rv;
406 } else {
407 // TODO: Move to a dynamic packet handler (like above)
408 if (s->read_buf_index >= s->read_buf_size)
409 return -1;
410 ret = mpegts_parse_packet(s->ts, pkt, s->buf + s->read_buf_index,
411 s->read_buf_size - s->read_buf_index);
412 if (ret < 0)
413 return -1;
414 s->read_buf_index += ret;
415 if (s->read_buf_index < s->read_buf_size)
416 return 1;
417 else
418 return 0;
419 }
420 }
421
422 if (len < 12)
423 return -1;
424
425 if ((buf[0] & 0xc0) != (RTP_VERSION << 6))
426 return -1;
427 if (buf[1] >= 200 && buf[1] <= 204) {
428 rtcp_parse_packet(s, buf, len);
429 return -1;
430 }
431 payload_type = buf[1] & 0x7f;
432 seq = AV_RB16(buf + 2);
433 timestamp = AV_RB32(buf + 4);
434 ssrc = AV_RB32(buf + 8);
435 /* store the ssrc in the RTPDemuxContext */
436 s->ssrc = ssrc;
437
438 /* NOTE: we can handle only one payload type */
439 if (s->payload_type != payload_type)
440 return -1;
441
442 st = s->st;
443 // only do something with this if all the rtp checks pass...
444 if(!rtp_valid_packet_in_sequence(&s->statistics, seq))
445 {
446 av_log(st?st->codec:NULL, AV_LOG_ERROR, "RTP: PT=%02x: bad cseq %04x expected=%04x\n",
447 payload_type, seq, ((s->seq + 1) & 0xffff));
448 return -1;
449 }
450
451 s->seq = seq;
452 len -= 12;
453 buf += 12;
454
455 if (!st) {
456 /* specific MPEG2TS demux support */
457 ret = mpegts_parse_packet(s->ts, pkt, buf, len);
458 if (ret < 0)
459 return -1;
460 if (ret < len) {
461 s->read_buf_size = len - ret;
462 memcpy(s->buf, buf + ret, s->read_buf_size);
463 s->read_buf_index = 0;
464 return 1;
465 }
b4e3330c 466 } else if (s->parse_packet) {
f841a0fc 467 rv = s->parse_packet(s, pkt, &timestamp, buf, len, flags);
8eb793c4
LA
468 } else {
469 // at this point, the RTP header has been stripped; This is ASSUMING that there is only 1 CSRC, which in't wise.
470 switch(st->codec->codec_id) {
471 case CODEC_ID_MP2:
472 /* better than nothing: skip mpeg audio RTP header */
473 if (len <= 4)
474 return -1;
475 h = AV_RB32(buf);
476 len -= 4;
477 buf += 4;
478 av_new_packet(pkt, len);
479 memcpy(pkt->data, buf, len);
480 break;
481 case CODEC_ID_MPEG1VIDEO:
482 case CODEC_ID_MPEG2VIDEO:
483 /* better than nothing: skip mpeg video RTP header */
484 if (len <= 4)
485 return -1;
486 h = AV_RB32(buf);
487 buf += 4;
488 len -= 4;
489 if (h & (1 << 26)) {
490 /* mpeg2 */
491 if (len <= 4)
492 return -1;
493 buf += 4;
494 len -= 4;
495 }
496 av_new_packet(pkt, len);
497 memcpy(pkt->data, buf, len);
498 break;
499 // moved from below, verbatim. this is because this section handles packets, and the lower switch handles
500 // timestamps.
501 // TODO: Put this into a dynamic packet handler...
502 case CODEC_ID_AAC:
503 if (rtp_parse_mp4_au(s, buf))
504 return -1;
505 {
506 rtp_payload_data_t *infos = s->rtp_payload_data;
507 if (infos == NULL)
508 return -1;
509 buf += infos->au_headers_length_bytes + 2;
510 len -= infos->au_headers_length_bytes + 2;
511
512 /* XXX: Fixme we only handle the case where rtp_parse_mp4_au define
513 one au_header */
514 av_new_packet(pkt, infos->au_headers[0].size);
515 memcpy(pkt->data, buf, infos->au_headers[0].size);
516 buf += infos->au_headers[0].size;
517 len -= infos->au_headers[0].size;
518 }
519 s->read_buf_size = len;
520 rv= 0;
521 break;
522 default:
f739b36d
RB
523 av_new_packet(pkt, len);
524 memcpy(pkt->data, buf, len);
8eb793c4
LA
525 break;
526 }
527
528 // now perform timestamp things....
529 finalize_packet(s, pkt, timestamp);
530 }
531 return rv;
532}
533
534void rtp_parse_close(RTPDemuxContext *s)
535{
536 // TODO: fold this into the protocol specific data fields.
537 if (!strcmp(ff_rtp_enc_name(s->payload_type), "MP2T")) {
538 mpegts_parse_close(s->ts);
539 }
540 av_free(s);
541}