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 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 "rtpdec.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/get_bits.h" 37 38struct 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 53RDTDemuxContext * 54ff_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 ? handler->parse_packet : NULL; 71 s->dynamic_protocol_context = priv_data; 72 73 return s; 74} 75 76void 77ff_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 87struct 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 96void 97ff_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, 1); 124 125 /* add tail */ 126 strcpy (response + 32, "01d0a8e3"); 127 128 /* calculate checksum */ 129 for (i = 0; i < 8; i++) 130 chksum[i] = response[i * 4]; 131 chksum[8] = 0; 132} 133 134static int 135rdt_load_mdpr (PayloadContext *rdt, AVStream *st, int rule_nr) 136{ 137 ByteIOContext pb; 138 int size; 139 uint32_t tag; 140 141 /** 142 * Layout of the MLTI chunk: 143 * 4:MLTI 144 * 2:<number of streams> 145 * Then for each stream ([number_of_streams] times): 146 * 2:<mdpr index> 147 * 2:<number of mdpr chunks> 148 * Then for each mdpr chunk ([number_of_mdpr_chunks] times): 149 * 4:<size> 150 * [size]:<data> 151 * we skip MDPR chunks until we reach the one of the stream 152 * we're interested in, and forward that ([size]+[data]) to 153 * the RM demuxer to parse the stream-specific header data. 154 */ 155 if (!rdt->mlti_data) 156 return -1; 157 init_put_byte(&pb, rdt->mlti_data, rdt->mlti_data_size, 0, 158 NULL, NULL, NULL, NULL); 159 tag = get_le32(&pb); 160 if (tag == MKTAG('M', 'L', 'T', 'I')) { 161 int num, chunk_nr; 162 163 /* read index of MDPR chunk numbers */ 164 num = get_be16(&pb); 165 if (rule_nr < 0 || rule_nr >= num) 166 return -1; 167 url_fskip(&pb, rule_nr * 2); 168 chunk_nr = get_be16(&pb); 169 url_fskip(&pb, (num - 1 - rule_nr) * 2); 170 171 /* read MDPR chunks */ 172 num = get_be16(&pb); 173 if (chunk_nr >= num) 174 return -1; 175 while (chunk_nr--) 176 url_fskip(&pb, get_be32(&pb)); 177 size = get_be32(&pb); 178 } else { 179 size = rdt->mlti_data_size; 180 url_fseek(&pb, 0, SEEK_SET); 181 } 182 if (ff_rm_read_mdpr_codecdata(rdt->rmctx, &pb, st, rdt->rmst[st->index], size) < 0) 183 return -1; 184 185 return 0; 186} 187 188/** 189 * Actual data handling. 190 */ 191 192int 193ff_rdt_parse_header(const uint8_t *buf, int len, 194 int *pset_id, int *pseq_no, int *pstream_id, 195 int *pis_keyframe, uint32_t *ptimestamp) 196{ 197 GetBitContext gb; 198 int consumed = 0, set_id, seq_no, stream_id, is_keyframe, 199 len_included, need_reliable; 200 uint32_t timestamp; 201 202 /* skip status packets */ 203 while (len >= 5 && buf[1] == 0xFF /* status packet */) { 204 int pkt_len; 205 206 if (!(buf[0] & 0x80)) 207 return -1; /* not followed by a data packet */ 208 209 pkt_len = AV_RB16(buf+3); 210 buf += pkt_len; 211 len -= pkt_len; 212 consumed += pkt_len; 213 } 214 if (len < 16) 215 return -1; 216 /** 217 * Layout of the header (in bits): 218 * 1: len_included 219 * Flag indicating whether this header includes a length field; 220 * this can be used to concatenate multiple RDT packets in a 221 * single UDP/TCP data frame and is used to precede RDT data 222 * by stream status packets 223 * 1: need_reliable 224 * Flag indicating whether this header includes a "reliable 225 * sequence number"; these are apparently sequence numbers of 226 * data packets alone. For data packets, this flag is always 227 * set, according to the Real documentation [1] 228 * 5: set_id 229 * ID of a set of streams of identical content, possibly with 230 * different codecs or bitrates 231 * 1: is_reliable 232 * Flag set for certain streams deemed less tolerable for packet 233 * loss 234 * 16: seq_no 235 * Packet sequence number; if >=0xFF00, this is a non-data packet 236 * containing stream status info, the second byte indicates the 237 * type of status packet (see wireshark docs / source code [2]) 238 * if (len_included) { 239 * 16: packet_len 240 * } else { 241 * packet_len = remainder of UDP/TCP frame 242 * } 243 * 1: is_back_to_back 244 * Back-to-Back flag; used for timing, set for one in every 10 245 * packets, according to the Real documentation [1] 246 * 1: is_slow_data 247 * Slow-data flag; currently unused, according to Real docs [1] 248 * 5: stream_id 249 * ID of the stream within this particular set of streams 250 * 1: is_no_keyframe 251 * Non-keyframe flag (unset if packet belongs to a keyframe) 252 * 32: timestamp (PTS) 253 * if (set_id == 0x1F) { 254 * 16: set_id (extended set-of-streams ID; see set_id) 255 * } 256 * if (need_reliable) { 257 * 16: reliable_seq_no 258 * Reliable sequence number (see need_reliable) 259 * } 260 * if (stream_id == 0x3F) { 261 * 16: stream_id (extended stream ID; see stream_id) 262 * } 263 * [1] https://protocol.helixcommunity.org/files/2005/devdocs/RDT_Feature_Level_20.txt 264 * [2] http://www.wireshark.org/docs/dfref/r/rdt.html and 265 * http://anonsvn.wireshark.org/viewvc/trunk/epan/dissectors/packet-rdt.c 266 */ 267 init_get_bits(&gb, buf, len << 3); 268 len_included = get_bits1(&gb); 269 need_reliable = get_bits1(&gb); 270 set_id = get_bits(&gb, 5); 271 skip_bits(&gb, 1); 272 seq_no = get_bits(&gb, 16); 273 if (len_included) 274 skip_bits(&gb, 16); 275 skip_bits(&gb, 2); 276 stream_id = get_bits(&gb, 5); 277 is_keyframe = !get_bits1(&gb); 278 timestamp = get_bits_long(&gb, 32); 279 if (set_id == 0x1f) 280 set_id = get_bits(&gb, 16); 281 if (need_reliable) 282 skip_bits(&gb, 16); 283 if (stream_id == 0x1f) 284 stream_id = get_bits(&gb, 16); 285 286 if (pset_id) *pset_id = set_id; 287 if (pseq_no) *pseq_no = seq_no; 288 if (pstream_id) *pstream_id = stream_id; 289 if (pis_keyframe) *pis_keyframe = is_keyframe; 290 if (ptimestamp) *ptimestamp = timestamp; 291 292 return consumed + (get_bits_count(&gb) >> 3); 293} 294 295/**< return 0 on packet, no more left, 1 on packet, 1 on partial packet... */ 296static int 297rdt_parse_packet (AVFormatContext *ctx, PayloadContext *rdt, AVStream *st, 298 AVPacket *pkt, uint32_t *timestamp, 299 const uint8_t *buf, int len, int flags) 300{ 301 int seq = 1, res; 302 ByteIOContext pb; 303 304 if (rdt->audio_pkt_cnt == 0) { 305 int pos; 306 307 init_put_byte(&pb, buf, len, 0, NULL, NULL, NULL, NULL); 308 flags = (flags & RTP_FLAG_KEY) ? 2 : 0; 309 res = ff_rm_parse_packet (rdt->rmctx, &pb, st, rdt->rmst[st->index], len, pkt, 310 &seq, flags, *timestamp); 311 pos = url_ftell(&pb); 312 if (res < 0) 313 return res; 314 if (res > 0) { 315 if (st->codec->codec_id == CODEC_ID_AAC) { 316 memcpy (rdt->buffer, buf + pos, len - pos); 317 rdt->rmctx->pb = av_alloc_put_byte (rdt->buffer, len - pos, 0, 318 NULL, NULL, NULL, NULL); 319 } 320 goto get_cache; 321 } 322 } else { 323get_cache: 324 rdt->audio_pkt_cnt = 325 ff_rm_retrieve_cache (rdt->rmctx, rdt->rmctx->pb, 326 st, rdt->rmst[st->index], pkt); 327 if (rdt->audio_pkt_cnt == 0 && 328 st->codec->codec_id == CODEC_ID_AAC) 329 av_freep(&rdt->rmctx->pb); 330 } 331 pkt->stream_index = st->index; 332 pkt->pts = *timestamp; 333 334 return rdt->audio_pkt_cnt > 0; 335} 336 337int 338ff_rdt_parse_packet(RDTDemuxContext *s, AVPacket *pkt, 339 const uint8_t *buf, int len) 340{ 341 int seq_no, flags = 0, stream_id, set_id, is_keyframe; 342 uint32_t timestamp; 343 int rv= 0; 344 345 if (!s->parse_packet) 346 return -1; 347 348 if (!buf && s->prev_stream_id != -1) { 349 /* return the next packets, if any */ 350 timestamp= 0; ///< Should not be used if buf is NULL, but should be set to the timestamp of the packet returned.... 351 rv= s->parse_packet(s->ic, s->dynamic_protocol_context, 352 s->streams[s->prev_stream_id], 353 pkt, ×tamp, NULL, 0, flags); 354 return rv; 355 } 356 357 if (len < 12) 358 return -1; 359 rv = ff_rdt_parse_header(buf, len, &set_id, &seq_no, &stream_id, &is_keyframe, ×tamp); 360 if (rv < 0) 361 return rv; 362 if (is_keyframe && 363 (set_id != s->prev_set_id || timestamp != s->prev_timestamp || 364 stream_id != s->prev_stream_id)) { 365 flags |= RTP_FLAG_KEY; 366 s->prev_set_id = set_id; 367 s->prev_timestamp = timestamp; 368 } 369 s->prev_stream_id = stream_id; 370 buf += rv; 371 len -= rv; 372 373 if (s->prev_stream_id >= s->n_streams) { 374 s->prev_stream_id = -1; 375 return -1; 376 } 377 378 rv = s->parse_packet(s->ic, s->dynamic_protocol_context, 379 s->streams[s->prev_stream_id], 380 pkt, ×tamp, buf, len, flags); 381 382 return rv; 383} 384 385void 386ff_rdt_subscribe_rule (char *cmd, int size, 387 int stream_nr, int rule_nr) 388{ 389 av_strlcatf(cmd, size, "stream=%d;rule=%d,stream=%d;rule=%d", 390 stream_nr, rule_nr * 2, stream_nr, rule_nr * 2 + 1); 391} 392 393static unsigned char * 394rdt_parse_b64buf (unsigned int *target_len, const char *p) 395{ 396 unsigned char *target; 397 int len = strlen(p); 398 if (*p == '\"') { 399 p++; 400 len -= 2; /* skip embracing " at start/end */ 401 } 402 *target_len = len * 3 / 4; 403 target = av_mallocz(*target_len + FF_INPUT_BUFFER_PADDING_SIZE); 404 av_base64_decode(target, p, *target_len); 405 return target; 406} 407 408static int 409rdt_parse_sdp_line (AVFormatContext *s, int st_index, 410 PayloadContext *rdt, const char *line) 411{ 412 AVStream *stream = s->streams[st_index]; 413 const char *p = line; 414 415 if (av_strstart(p, "OpaqueData:buffer;", &p)) { 416 rdt->mlti_data = rdt_parse_b64buf(&rdt->mlti_data_size, p); 417 } else if (av_strstart(p, "StartTime:integer;", &p)) 418 stream->first_dts = atoi(p); 419 else if (av_strstart(p, "ASMRuleBook:string;", &p)) { 420 int n, first = -1; 421 422 for (n = 0; n < s->nb_streams; n++) 423 if (s->streams[n]->priv_data == stream->priv_data) { 424 if (first == -1) first = n; 425 rdt->rmst[s->streams[n]->index] = ff_rm_alloc_rmstream(); 426 rdt_load_mdpr(rdt, s->streams[n], (n - first) * 2); 427 428 if (s->streams[n]->codec->codec_id == CODEC_ID_AAC) 429 s->streams[n]->codec->frame_size = 1; // FIXME 430 } 431 } 432 433 return 0; 434} 435 436static void 437real_parse_asm_rule(AVStream *st, const char *p, const char *end) 438{ 439 do { 440 /* can be either averagebandwidth= or AverageBandwidth= */ 441 if (sscanf(p, " %*1[Aa]verage%*1[Bb]andwidth=%d", &st->codec->bit_rate) == 1) 442 break; 443 if (!(p = strchr(p, ',')) || p > end) 444 p = end; 445 p++; 446 } while (p < end); 447} 448 449static AVStream * 450add_dstream(AVFormatContext *s, AVStream *orig_st) 451{ 452 AVStream *st; 453 454 if (!(st = av_new_stream(s, 0))) 455 return NULL; 456 st->codec->codec_type = orig_st->codec->codec_type; 457 st->priv_data = orig_st->priv_data; 458 st->first_dts = orig_st->first_dts; 459 460 return st; 461} 462 463static void 464real_parse_asm_rulebook(AVFormatContext *s, AVStream *orig_st, 465 const char *p) 466{ 467 const char *end; 468 int n_rules, odd = 0; 469 AVStream *st; 470 471 /** 472 * The ASMRuleBook contains a list of comma-separated strings per rule, 473 * and each rule is separated by a ;. The last one also has a ; at the 474 * end so we can use it as delimiter. 475 * Every rule occurs twice, once for when the RTSP packet header marker 476 * is set and once for if it isn't. We only read the first because we 477 * don't care much (that's what the "odd" variable is for). 478 * Each rule contains a set of one or more statements, optionally 479 * preceeded by a single condition. If there's a condition, the rule 480 * starts with a '#'. Multiple conditions are merged between brackets, 481 * so there are never multiple conditions spread out over separate 482 * statements. Generally, these conditions are bitrate limits (min/max) 483 * for multi-bitrate streams. 484 */ 485 if (*p == '\"') p++; 486 for (n_rules = 0; s->nb_streams < MAX_STREAMS;) { 487 if (!(end = strchr(p, ';'))) 488 break; 489 if (!odd && end != p) { 490 if (n_rules > 0) 491 st = add_dstream(s, orig_st); 492 else 493 st = orig_st; 494 real_parse_asm_rule(st, p, end); 495 n_rules++; 496 } 497 p = end + 1; 498 odd ^= 1; 499 } 500} 501 502void 503ff_real_parse_sdp_a_line (AVFormatContext *s, int stream_index, 504 const char *line) 505{ 506 const char *p = line; 507 508 if (av_strstart(p, "ASMRuleBook:string;", &p)) 509 real_parse_asm_rulebook(s, s->streams[stream_index], p); 510} 511 512static PayloadContext * 513rdt_new_context (void) 514{ 515 PayloadContext *rdt = av_mallocz(sizeof(PayloadContext)); 516 517 av_open_input_stream(&rdt->rmctx, NULL, "", &rdt_demuxer, NULL); 518 519 return rdt; 520} 521 522static void 523rdt_free_context (PayloadContext *rdt) 524{ 525 int i; 526 527 for (i = 0; i < MAX_STREAMS; i++) 528 if (rdt->rmst[i]) { 529 ff_rm_free_rmstream(rdt->rmst[i]); 530 av_freep(&rdt->rmst[i]); 531 } 532 if (rdt->rmctx) 533 av_close_input_stream(rdt->rmctx); 534 av_freep(&rdt->mlti_data); 535 av_free(rdt); 536} 537 538#define RDT_HANDLER(n, s, t) \ 539static RTPDynamicProtocolHandler ff_rdt_ ## n ## _handler = { \ 540 .enc_name = s, \ 541 .codec_type = t, \ 542 .codec_id = CODEC_ID_NONE, \ 543 .parse_sdp_a_line = rdt_parse_sdp_line, \ 544 .open = rdt_new_context, \ 545 .close = rdt_free_context, \ 546 .parse_packet = rdt_parse_packet \ 547}; 548 549RDT_HANDLER(live_video, "x-pn-multirate-realvideo-live", AVMEDIA_TYPE_VIDEO); 550RDT_HANDLER(live_audio, "x-pn-multirate-realaudio-live", AVMEDIA_TYPE_AUDIO); 551RDT_HANDLER(video, "x-pn-realvideo", AVMEDIA_TYPE_VIDEO); 552RDT_HANDLER(audio, "x-pn-realaudio", AVMEDIA_TYPE_AUDIO); 553 554void av_register_rdt_dynamic_payload_handlers(void) 555{ 556 ff_register_dynamic_payload_handler(&ff_rdt_video_handler); 557 ff_register_dynamic_payload_handler(&ff_rdt_audio_handler); 558 ff_register_dynamic_payload_handler(&ff_rdt_live_video_handler); 559 ff_register_dynamic_payload_handler(&ff_rdt_live_audio_handler); 560} 561