Merge rtp_internal.h in rtp.h, and remove rtp_internal.h
[libav.git] / libavformat / rdt.c
1 /*
2 * Realmedia RTSP protocol (RDT) support.
3 * Copyright (c) 2007 Ronald S. Bultje
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 /**
23 * @file rdt.c
24 * @brief Realmedia RTSP protocol (RDT) support
25 * @author Ronald S. Bultje <rbultje@ronald.bitfreak.net>
26 */
27
28 #include "avformat.h"
29 #include "libavutil/avstring.h"
30 #include "rtp.h"
31 #include "rdt.h"
32 #include "libavutil/base64.h"
33 #include "libavutil/md5.h"
34 #include "rm.h"
35 #include "internal.h"
36 #include "libavcodec/bitstream.h"
37
38 struct RDTDemuxContext {
39 AVFormatContext *ic; /**< the containing (RTSP) demux context */
40 /** Each RDT stream-set (represented by one RTSPStream) can contain
41 * multiple streams (of the same content, but with possibly different
42 * codecs/bitrates). Each such stream is represented by one AVStream
43 * in the AVFormatContext, and this variable points to the offset in
44 * that array such that the first is the first stream of this set. */
45 AVStream **streams;
46 int n_streams; /**< streams with identifical content in this set */
47 void *dynamic_protocol_context;
48 DynamicPayloadPacketHandlerProc parse_packet;
49 uint32_t prev_timestamp;
50 int prev_set_id, prev_stream_id;
51 };
52
53 RDTDemuxContext *
54 ff_rdt_parse_open(AVFormatContext *ic, int first_stream_of_set_idx,
55 void *priv_data, RTPDynamicProtocolHandler *handler)
56 {
57 RDTDemuxContext *s = av_mallocz(sizeof(RDTDemuxContext));
58 if (!s)
59 return NULL;
60
61 s->ic = ic;
62 s->streams = &ic->streams[first_stream_of_set_idx];
63 do {
64 s->n_streams++;
65 } while (first_stream_of_set_idx + s->n_streams < ic->nb_streams &&
66 s->streams[s->n_streams]->priv_data == s->streams[0]->priv_data);
67 s->prev_set_id = -1;
68 s->prev_stream_id = -1;
69 s->prev_timestamp = -1;
70 s->parse_packet = handler->parse_packet;
71 s->dynamic_protocol_context = priv_data;
72
73 return s;
74 }
75
76 void
77 ff_rdt_parse_close(RDTDemuxContext *s)
78 {
79 int i;
80
81 for (i = 1; i < s->n_streams; i++)
82 s->streams[i]->priv_data = NULL;
83
84 av_free(s);
85 }
86
87 struct PayloadContext {
88 AVFormatContext *rmctx;
89 RMStream *rmst[MAX_STREAMS];
90 uint8_t *mlti_data;
91 unsigned int mlti_data_size;
92 char buffer[RTP_MAX_PACKET_LENGTH + FF_INPUT_BUFFER_PADDING_SIZE];
93 int audio_pkt_cnt; /**< remaining audio packets in rmdec */
94 };
95
96 void
97 ff_rdt_calc_response_and_checksum(char response[41], char chksum[9],
98 const char *challenge)
99 {
100 int ch_len = strlen (challenge), i;
101 unsigned char zres[16],
102 buf[64] = { 0xa1, 0xe9, 0x14, 0x9d, 0x0e, 0x6b, 0x3b, 0x59 };
103 #define XOR_TABLE_SIZE 37
104 const unsigned char xor_table[XOR_TABLE_SIZE] = {
105 0x05, 0x18, 0x74, 0xd0, 0x0d, 0x09, 0x02, 0x53,
106 0xc0, 0x01, 0x05, 0x05, 0x67, 0x03, 0x19, 0x70,
107 0x08, 0x27, 0x66, 0x10, 0x10, 0x72, 0x08, 0x09,
108 0x63, 0x11, 0x03, 0x71, 0x08, 0x08, 0x70, 0x02,
109 0x10, 0x57, 0x05, 0x18, 0x54 };
110
111 /* some (length) checks */
112 if (ch_len == 40) /* what a hack... */
113 ch_len = 32;
114 else if (ch_len > 56)
115 ch_len = 56;
116 memcpy(buf + 8, challenge, ch_len);
117
118 /* xor challenge bytewise with xor_table */
119 for (i = 0; i < XOR_TABLE_SIZE; i++)
120 buf[8 + i] ^= xor_table[i];
121
122 av_md5_sum(zres, buf, 64);
123 ff_data_to_hex(response, zres, 16);
124 for (i=0;i<32;i++) response[i] = tolower(response[i]);
125
126 /* add tail */
127 strcpy (response + 32, "01d0a8e3");
128
129 /* calculate checksum */
130 for (i = 0; i < 8; i++)
131 chksum[i] = response[i * 4];
132 chksum[8] = 0;
133 }
134
135 static int
136 rdt_load_mdpr (PayloadContext *rdt, AVStream *st, int rule_nr)
137 {
138 ByteIOContext pb;
139 int size;
140 uint32_t tag;
141
142 /**
143 * Layout of the MLTI chunk:
144 * 4:MLTI
145 * 2:<number of streams>
146 * Then for each stream ([number_of_streams] times):
147 * 2:<mdpr index>
148 * 2:<number of mdpr chunks>
149 * Then for each mdpr chunk ([number_of_mdpr_chunks] times):
150 * 4:<size>
151 * [size]:<data>
152 * we skip MDPR chunks until we reach the one of the stream
153 * we're interested in, and forward that ([size]+[data]) to
154 * the RM demuxer to parse the stream-specific header data.
155 */
156 if (!rdt->mlti_data)
157 return -1;
158 init_put_byte(&pb, rdt->mlti_data, rdt->mlti_data_size, 0,
159 NULL, NULL, NULL, NULL);
160 tag = get_le32(&pb);
161 if (tag == MKTAG('M', 'L', 'T', 'I')) {
162 int num, chunk_nr;
163
164 /* read index of MDPR chunk numbers */
165 num = get_be16(&pb);
166 if (rule_nr < 0 || rule_nr >= num)
167 return -1;
168 url_fskip(&pb, rule_nr * 2);
169 chunk_nr = get_be16(&pb);
170 url_fskip(&pb, (num - 1 - rule_nr) * 2);
171
172 /* read MDPR chunks */
173 num = get_be16(&pb);
174 if (chunk_nr >= num)
175 return -1;
176 while (chunk_nr--)
177 url_fskip(&pb, get_be32(&pb));
178 size = get_be32(&pb);
179 } else {
180 size = rdt->mlti_data_size;
181 url_fseek(&pb, 0, SEEK_SET);
182 }
183 if (ff_rm_read_mdpr_codecdata(rdt->rmctx, &pb, st, rdt->rmst[st->index], size) < 0)
184 return -1;
185
186 return 0;
187 }
188
189 /**
190 * Actual data handling.
191 */
192
193 int
194 ff_rdt_parse_header(const uint8_t *buf, int len,
195 int *pset_id, int *pseq_no, int *pstream_id,
196 int *pis_keyframe, uint32_t *ptimestamp)
197 {
198 GetBitContext gb;
199 int consumed = 0, set_id, seq_no, stream_id, is_keyframe,
200 len_included, need_reliable;
201 uint32_t timestamp;
202
203 /* skip status packets */
204 while (len >= 5 && buf[1] == 0xFF /* status packet */) {
205 int pkt_len;
206
207 if (!(buf[0] & 0x80))
208 return -1; /* not followed by a data packet */
209
210 pkt_len = AV_RB16(buf+3);
211 buf += pkt_len;
212 len -= pkt_len;
213 consumed += pkt_len;
214 }
215 if (len < 16)
216 return -1;
217 /**
218 * Layout of the header (in bits):
219 * 1: len_included
220 * Flag indicating whether this header includes a length field;
221 * this can be used to concatenate multiple RDT packets in a
222 * single UDP/TCP data frame and is used to precede RDT data
223 * by stream status packets
224 * 1: need_reliable
225 * Flag indicating whether this header includes a "reliable
226 * sequence number"; these are apparently sequence numbers of
227 * data packets alone. For data packets, this flag is always
228 * set, according to the Real documentation [1]
229 * 5: set_id
230 * ID of a set of streams of identical content, possibly with
231 * different codecs or bitrates
232 * 1: is_reliable
233 * Flag set for certain streams deemed less tolerable for packet
234 * loss
235 * 16: seq_no
236 * Packet sequence number; if >=0xFF00, this is a non-data packet
237 * containing stream status info, the second byte indicates the
238 * type of status packet (see wireshark docs / source code [2])
239 * if (len_included) {
240 * 16: packet_len
241 * } else {
242 * packet_len = remainder of UDP/TCP frame
243 * }
244 * 1: is_back_to_back
245 * Back-to-Back flag; used for timing, set for one in every 10
246 * packets, according to the Real documentation [1]
247 * 1: is_slow_data
248 * Slow-data flag; currently unused, according to Real docs [1]
249 * 5: stream_id
250 * ID of the stream within this particular set of streams
251 * 1: is_no_keyframe
252 * Non-keyframe flag (unset if packet belongs to a keyframe)
253 * 32: timestamp (PTS)
254 * if (set_id == 0x1F) {
255 * 16: set_id (extended set-of-streams ID; see set_id)
256 * }
257 * if (need_reliable) {
258 * 16: reliable_seq_no
259 * Reliable sequence number (see need_reliable)
260 * }
261 * if (stream_id == 0x3F) {
262 * 16: stream_id (extended stream ID; see stream_id)
263 * }
264 * [1] https://protocol.helixcommunity.org/files/2005/devdocs/RDT_Feature_Level_20.txt
265 * [2] http://www.wireshark.org/docs/dfref/r/rdt.html and
266 * http://anonsvn.wireshark.org/viewvc/trunk/epan/dissectors/packet-rdt.c
267 */
268 init_get_bits(&gb, buf, len << 3);
269 len_included = get_bits1(&gb);
270 need_reliable = get_bits1(&gb);
271 set_id = get_bits(&gb, 5);
272 skip_bits(&gb, 1);
273 seq_no = get_bits(&gb, 16);
274 if (len_included)
275 skip_bits(&gb, 16);
276 skip_bits(&gb, 2);
277 stream_id = get_bits(&gb, 5);
278 is_keyframe = !get_bits1(&gb);
279 timestamp = get_bits_long(&gb, 32);
280 if (set_id == 0x1f)
281 set_id = get_bits(&gb, 16);
282 if (need_reliable)
283 skip_bits(&gb, 16);
284 if (stream_id == 0x1f)
285 stream_id = get_bits(&gb, 16);
286
287 if (pset_id) *pset_id = set_id;
288 if (pseq_no) *pseq_no = seq_no;
289 if (pstream_id) *pstream_id = stream_id;
290 if (pis_keyframe) *pis_keyframe = is_keyframe;
291 if (ptimestamp) *ptimestamp = timestamp;
292
293 return consumed + (get_bits_count(&gb) >> 3);
294 }
295
296 /**< return 0 on packet, no more left, 1 on packet, 1 on partial packet... */
297 static int
298 rdt_parse_packet (PayloadContext *rdt, AVStream *st,
299 AVPacket *pkt, uint32_t *timestamp,
300 const uint8_t *buf, int len, int flags)
301 {
302 int seq = 1, res;
303 ByteIOContext pb;
304
305 if (rdt->audio_pkt_cnt == 0) {
306 int pos;
307
308 init_put_byte(&pb, buf, len, 0, NULL, NULL, NULL, NULL);
309 flags = (flags & PKT_FLAG_KEY) ? 2 : 0;
310 res = ff_rm_parse_packet (rdt->rmctx, &pb, st, rdt->rmst[st->index], len, pkt,
311 &seq, &flags, timestamp);
312 pos = url_ftell(&pb);
313 if (res < 0)
314 return res;
315 rdt->audio_pkt_cnt = res;
316 if (rdt->audio_pkt_cnt > 0 &&
317 st->codec->codec_id == CODEC_ID_AAC) {
318 memcpy (rdt->buffer, buf + pos, len - pos);
319 rdt->rmctx->pb = av_alloc_put_byte (rdt->buffer, len - pos, 0,
320 NULL, NULL, NULL, NULL);
321 }
322 } else {
323 rdt->audio_pkt_cnt =
324 ff_rm_retrieve_cache (rdt->rmctx, rdt->rmctx->pb,
325 st, rdt->rmst[st->index], pkt);
326 if (rdt->audio_pkt_cnt == 0 &&
327 st->codec->codec_id == CODEC_ID_AAC)
328 av_freep(&rdt->rmctx->pb);
329 }
330 pkt->stream_index = st->index;
331 pkt->pts = *timestamp;
332
333 return rdt->audio_pkt_cnt > 0;
334 }
335
336 int
337 ff_rdt_parse_packet(RDTDemuxContext *s, AVPacket *pkt,
338 const uint8_t *buf, int len)
339 {
340 int seq_no, flags = 0, stream_id, set_id, is_keyframe;
341 uint32_t timestamp;
342 int rv= 0;
343
344 if (!s->parse_packet)
345 return -1;
346
347 if (!buf && s->prev_stream_id != -1) {
348 /* return the next packets, if any */
349 timestamp= 0; ///< Should not be used if buf is NULL, but should be set to the timestamp of the packet returned....
350 rv= s->parse_packet(s->dynamic_protocol_context,
351 s->streams[s->prev_stream_id],
352 pkt, &timestamp, NULL, 0, flags);
353 return rv;
354 }
355
356 if (len < 12)
357 return -1;
358 rv = ff_rdt_parse_header(buf, len, &set_id, &seq_no, &stream_id, &is_keyframe, &timestamp);
359 if (rv < 0)
360 return rv;
361 if (is_keyframe &&
362 (set_id != s->prev_set_id || timestamp != s->prev_timestamp ||
363 stream_id != s->prev_stream_id)) {
364 flags |= PKT_FLAG_KEY;
365 s->prev_set_id = set_id;
366 s->prev_timestamp = timestamp;
367 }
368 s->prev_stream_id = stream_id;
369 buf += rv;
370 len -= rv;
371
372 if (s->prev_stream_id >= s->n_streams) {
373 s->prev_stream_id = -1;
374 return -1;
375 }
376
377 rv = s->parse_packet(s->dynamic_protocol_context,
378 s->streams[s->prev_stream_id],
379 pkt, &timestamp, buf, len, flags);
380
381 return rv;
382 }
383
384 void
385 ff_rdt_subscribe_rule (char *cmd, int size,
386 int stream_nr, int rule_nr)
387 {
388 av_strlcatf(cmd, size, "stream=%d;rule=%d,stream=%d;rule=%d",
389 stream_nr, rule_nr * 2, stream_nr, rule_nr * 2 + 1);
390 }
391
392 static unsigned char *
393 rdt_parse_b64buf (unsigned int *target_len, const char *p)
394 {
395 unsigned char *target;
396 int len = strlen(p);
397 if (*p == '\"') {
398 p++;
399 len -= 2; /* skip embracing " at start/end */
400 }
401 *target_len = len * 3 / 4;
402 target = av_mallocz(*target_len + FF_INPUT_BUFFER_PADDING_SIZE);
403 av_base64_decode(target, p, *target_len);
404 return target;
405 }
406
407 static int
408 rdt_parse_sdp_line (AVFormatContext *s, int st_index,
409 PayloadContext *rdt, const char *line)
410 {
411 AVStream *stream = s->streams[st_index];
412 const char *p = line;
413
414 if (av_strstart(p, "OpaqueData:buffer;", &p)) {
415 rdt->mlti_data = rdt_parse_b64buf(&rdt->mlti_data_size, p);
416 } else if (av_strstart(p, "StartTime:integer;", &p))
417 stream->first_dts = atoi(p);
418 else if (av_strstart(p, "ASMRuleBook:string;", &p)) {
419 int n = st_index, first = -1;
420
421 for (n = 0; n < s->nb_streams; n++)
422 if (s->streams[n]->priv_data == stream->priv_data) {
423 if (first == -1) first = n;
424 rdt->rmst[s->streams[n]->index] = ff_rm_alloc_rmstream();
425 rdt_load_mdpr(rdt, s->streams[n], (n - first) * 2);
426
427 if (s->streams[n]->codec->codec_id == CODEC_ID_AAC)
428 s->streams[n]->codec->frame_size = 1; // FIXME
429 }
430 }
431
432 return 0;
433 }
434
435 static void
436 real_parse_asm_rule(AVStream *st, const char *p, const char *end)
437 {
438 do {
439 /* can be either averagebandwidth= or AverageBandwidth= */
440 if (sscanf(p, " %*1[Aa]verage%*1[Bb]andwidth=%d", &st->codec->bit_rate) == 1)
441 break;
442 if (!(p = strchr(p, ',')) || p > end)
443 p = end;
444 p++;
445 } while (p < end);
446 }
447
448 static AVStream *
449 add_dstream(AVFormatContext *s, AVStream *orig_st)
450 {
451 AVStream *st;
452
453 if (!(st = av_new_stream(s, 0)))
454 return NULL;
455 st->codec->codec_type = orig_st->codec->codec_type;
456 st->priv_data = orig_st->priv_data;
457 st->first_dts = orig_st->first_dts;
458
459 return st;
460 }
461
462 static void
463 real_parse_asm_rulebook(AVFormatContext *s, AVStream *orig_st,
464 const char *p)
465 {
466 const char *end;
467 int n_rules, odd = 0;
468 AVStream *st;
469
470 /**
471 * The ASMRuleBook contains a list of comma-separated strings per rule,
472 * and each rule is separated by a ;. The last one also has a ; at the
473 * end so we can use it as delimiter.
474 * Every rule occurs twice, once for when the RTSP packet header marker
475 * is set and once for if it isn't. We only read the first because we
476 * don't care much (that's what the "odd" variable is for).
477 * Each rule contains a set of one or more statements, optionally
478 * preceeded by a single condition. If there's a condition, the rule
479 * starts with a '#'. Multiple conditions are merged between brackets,
480 * so there are never multiple conditions spread out over separate
481 * statements. Generally, these conditions are bitrate limits (min/max)
482 * for multi-bitrate streams.
483 */
484 if (*p == '\"') p++;
485 for (n_rules = 0; s->nb_streams < MAX_STREAMS;) {
486 if (!(end = strchr(p, ';')))
487 break;
488 if (!odd && end != p) {
489 if (n_rules > 0)
490 st = add_dstream(s, orig_st);
491 else
492 st = orig_st;
493 real_parse_asm_rule(st, p, end);
494 n_rules++;
495 }
496 p = end + 1;
497 odd ^= 1;
498 }
499 }
500
501 void
502 ff_real_parse_sdp_a_line (AVFormatContext *s, int stream_index,
503 const char *line)
504 {
505 const char *p = line;
506
507 if (av_strstart(p, "ASMRuleBook:string;", &p))
508 real_parse_asm_rulebook(s, s->streams[stream_index], p);
509 }
510
511 static PayloadContext *
512 rdt_new_extradata (void)
513 {
514 PayloadContext *rdt = av_mallocz(sizeof(PayloadContext));
515
516 av_open_input_stream(&rdt->rmctx, NULL, "", &rdt_demuxer, NULL);
517
518 return rdt;
519 }
520
521 static void
522 rdt_free_extradata (PayloadContext *rdt)
523 {
524 int i;
525
526 for (i = 0; i < MAX_STREAMS; i++)
527 if (rdt->rmst[i]) {
528 ff_rm_free_rmstream(rdt->rmst[i]);
529 av_freep(&rdt->rmst[i]);
530 }
531 if (rdt->rmctx)
532 av_close_input_stream(rdt->rmctx);
533 av_freep(&rdt->mlti_data);
534 av_free(rdt);
535 }
536
537 #define RDT_HANDLER(n, s, t) \
538 static RTPDynamicProtocolHandler ff_rdt_ ## n ## _handler = { \
539 s, \
540 t, \
541 CODEC_ID_NONE, \
542 rdt_parse_sdp_line, \
543 rdt_new_extradata, \
544 rdt_free_extradata, \
545 rdt_parse_packet \
546 };
547
548 RDT_HANDLER(live_video, "x-pn-multirate-realvideo-live", CODEC_TYPE_VIDEO);
549 RDT_HANDLER(live_audio, "x-pn-multirate-realaudio-live", CODEC_TYPE_AUDIO);
550 RDT_HANDLER(video, "x-pn-realvideo", CODEC_TYPE_VIDEO);
551 RDT_HANDLER(audio, "x-pn-realaudio", CODEC_TYPE_AUDIO);
552
553 void av_register_rdt_dynamic_payload_handlers(void)
554 {
555 ff_register_dynamic_payload_handler(&ff_rdt_video_handler);
556 ff_register_dynamic_payload_handler(&ff_rdt_audio_handler);
557 ff_register_dynamic_payload_handler(&ff_rdt_live_video_handler);
558 ff_register_dynamic_payload_handler(&ff_rdt_live_audio_handler);
559 }