| 1 | /* |
| 2 | * Core Audio Format demuxer |
| 3 | * Copyright (c) 2007 Justin Ruggles |
| 4 | * Copyright (c) 2009 Peter Ross |
| 5 | * |
| 6 | * This file is part of FFmpeg. |
| 7 | * |
| 8 | * FFmpeg is free software; you can redistribute it and/or |
| 9 | * modify it under the terms of the GNU Lesser General Public |
| 10 | * License as published by the Free Software Foundation; either |
| 11 | * version 2.1 of the License, or (at your option) any later version. |
| 12 | * |
| 13 | * FFmpeg is distributed in the hope that it will be useful, |
| 14 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 15 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 16 | * Lesser General Public License for more details. |
| 17 | * |
| 18 | * You should have received a copy of the GNU Lesser General Public |
| 19 | * License along with FFmpeg; if not, write to the Free Software |
| 20 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 21 | */ |
| 22 | |
| 23 | /** |
| 24 | * @file |
| 25 | * Core Audio Format demuxer |
| 26 | */ |
| 27 | |
| 28 | #include <inttypes.h> |
| 29 | |
| 30 | #include "avformat.h" |
| 31 | #include "internal.h" |
| 32 | #include "isom.h" |
| 33 | #include "mov_chan.h" |
| 34 | #include "libavutil/intreadwrite.h" |
| 35 | #include "libavutil/intfloat.h" |
| 36 | #include "libavutil/dict.h" |
| 37 | #include "caf.h" |
| 38 | |
| 39 | typedef struct { |
| 40 | int bytes_per_packet; ///< bytes in a packet, or 0 if variable |
| 41 | int frames_per_packet; ///< frames in a packet, or 0 if variable |
| 42 | int64_t num_bytes; ///< total number of bytes in stream |
| 43 | |
| 44 | int64_t packet_cnt; ///< packet counter |
| 45 | int64_t frame_cnt; ///< frame counter |
| 46 | |
| 47 | int64_t data_start; ///< data start position, in bytes |
| 48 | int64_t data_size; ///< raw data size, in bytes |
| 49 | } CafContext; |
| 50 | |
| 51 | static int probe(AVProbeData *p) |
| 52 | { |
| 53 | if (AV_RB32(p->buf) == MKBETAG('c','a','f','f') && AV_RB16(&p->buf[4]) == 1) |
| 54 | return AVPROBE_SCORE_MAX; |
| 55 | return 0; |
| 56 | } |
| 57 | |
| 58 | /** Read audio description chunk */ |
| 59 | static int read_desc_chunk(AVFormatContext *s) |
| 60 | { |
| 61 | AVIOContext *pb = s->pb; |
| 62 | CafContext *caf = s->priv_data; |
| 63 | AVStream *st; |
| 64 | int flags; |
| 65 | |
| 66 | /* new audio stream */ |
| 67 | st = avformat_new_stream(s, NULL); |
| 68 | if (!st) |
| 69 | return AVERROR(ENOMEM); |
| 70 | |
| 71 | /* parse format description */ |
| 72 | st->codec->codec_type = AVMEDIA_TYPE_AUDIO; |
| 73 | st->codec->sample_rate = av_int2double(avio_rb64(pb)); |
| 74 | st->codec->codec_tag = avio_rl32(pb); |
| 75 | flags = avio_rb32(pb); |
| 76 | caf->bytes_per_packet = avio_rb32(pb); |
| 77 | st->codec->block_align = caf->bytes_per_packet; |
| 78 | caf->frames_per_packet = avio_rb32(pb); |
| 79 | st->codec->channels = avio_rb32(pb); |
| 80 | st->codec->bits_per_coded_sample = avio_rb32(pb); |
| 81 | |
| 82 | /* calculate bit rate for constant size packets */ |
| 83 | if (caf->frames_per_packet > 0 && caf->bytes_per_packet > 0) { |
| 84 | st->codec->bit_rate = (uint64_t)st->codec->sample_rate * (uint64_t)caf->bytes_per_packet * 8 |
| 85 | / (uint64_t)caf->frames_per_packet; |
| 86 | } else { |
| 87 | st->codec->bit_rate = 0; |
| 88 | } |
| 89 | |
| 90 | /* determine codec */ |
| 91 | if (st->codec->codec_tag == MKTAG('l','p','c','m')) |
| 92 | st->codec->codec_id = ff_mov_get_lpcm_codec_id(st->codec->bits_per_coded_sample, (flags ^ 0x2) | 0x4); |
| 93 | else |
| 94 | st->codec->codec_id = ff_codec_get_id(ff_codec_caf_tags, st->codec->codec_tag); |
| 95 | return 0; |
| 96 | } |
| 97 | |
| 98 | /** Read magic cookie chunk */ |
| 99 | static int read_kuki_chunk(AVFormatContext *s, int64_t size) |
| 100 | { |
| 101 | AVIOContext *pb = s->pb; |
| 102 | AVStream *st = s->streams[0]; |
| 103 | |
| 104 | if (size < 0 || size > INT_MAX - FF_INPUT_BUFFER_PADDING_SIZE) |
| 105 | return -1; |
| 106 | |
| 107 | if (st->codec->codec_id == AV_CODEC_ID_AAC) { |
| 108 | /* The magic cookie format for AAC is an mp4 esds atom. |
| 109 | The lavc AAC decoder requires the data from the codec specific |
| 110 | description as extradata input. */ |
| 111 | int strt, skip; |
| 112 | |
| 113 | strt = avio_tell(pb); |
| 114 | ff_mov_read_esds(s, pb); |
| 115 | skip = size - (avio_tell(pb) - strt); |
| 116 | if (skip < 0 || !st->codec->extradata || |
| 117 | st->codec->codec_id != AV_CODEC_ID_AAC) { |
| 118 | av_log(s, AV_LOG_ERROR, "invalid AAC magic cookie\n"); |
| 119 | return AVERROR_INVALIDDATA; |
| 120 | } |
| 121 | avio_skip(pb, skip); |
| 122 | } else if (st->codec->codec_id == AV_CODEC_ID_ALAC) { |
| 123 | #define ALAC_PREAMBLE 12 |
| 124 | #define ALAC_HEADER 36 |
| 125 | #define ALAC_NEW_KUKI 24 |
| 126 | uint8_t preamble[12]; |
| 127 | if (size < ALAC_NEW_KUKI) { |
| 128 | av_log(s, AV_LOG_ERROR, "invalid ALAC magic cookie\n"); |
| 129 | avio_skip(pb, size); |
| 130 | return AVERROR_INVALIDDATA; |
| 131 | } |
| 132 | avio_read(pb, preamble, ALAC_PREAMBLE); |
| 133 | |
| 134 | if (ff_alloc_extradata(st->codec, ALAC_HEADER)) |
| 135 | return AVERROR(ENOMEM); |
| 136 | |
| 137 | /* For the old style cookie, we skip 12 bytes, then read 36 bytes. |
| 138 | * The new style cookie only contains the last 24 bytes of what was |
| 139 | * 36 bytes in the old style cookie, so we fabricate the first 12 bytes |
| 140 | * in that case to maintain compatibility. */ |
| 141 | if (!memcmp(&preamble[4], "frmaalac", 8)) { |
| 142 | if (size < ALAC_PREAMBLE + ALAC_HEADER) { |
| 143 | av_log(s, AV_LOG_ERROR, "invalid ALAC magic cookie\n"); |
| 144 | av_freep(&st->codec->extradata); |
| 145 | return AVERROR_INVALIDDATA; |
| 146 | } |
| 147 | avio_read(pb, st->codec->extradata, ALAC_HEADER); |
| 148 | avio_skip(pb, size - ALAC_PREAMBLE - ALAC_HEADER); |
| 149 | } else { |
| 150 | AV_WB32(st->codec->extradata, 36); |
| 151 | memcpy(&st->codec->extradata[4], "alac", 4); |
| 152 | AV_WB32(&st->codec->extradata[8], 0); |
| 153 | memcpy(&st->codec->extradata[12], preamble, 12); |
| 154 | avio_read(pb, &st->codec->extradata[24], ALAC_NEW_KUKI - 12); |
| 155 | avio_skip(pb, size - ALAC_NEW_KUKI); |
| 156 | } |
| 157 | } else { |
| 158 | if (ff_get_extradata(st->codec, pb, size) < 0) |
| 159 | return AVERROR(ENOMEM); |
| 160 | } |
| 161 | |
| 162 | return 0; |
| 163 | } |
| 164 | |
| 165 | /** Read packet table chunk */ |
| 166 | static int read_pakt_chunk(AVFormatContext *s, int64_t size) |
| 167 | { |
| 168 | AVIOContext *pb = s->pb; |
| 169 | AVStream *st = s->streams[0]; |
| 170 | CafContext *caf = s->priv_data; |
| 171 | int64_t pos = 0, ccount, num_packets; |
| 172 | int i; |
| 173 | |
| 174 | ccount = avio_tell(pb); |
| 175 | |
| 176 | num_packets = avio_rb64(pb); |
| 177 | if (num_packets < 0 || INT32_MAX / sizeof(AVIndexEntry) < num_packets) |
| 178 | return AVERROR_INVALIDDATA; |
| 179 | |
| 180 | st->nb_frames = avio_rb64(pb); /* valid frames */ |
| 181 | st->nb_frames += avio_rb32(pb); /* priming frames */ |
| 182 | st->nb_frames += avio_rb32(pb); /* remainder frames */ |
| 183 | |
| 184 | st->duration = 0; |
| 185 | for (i = 0; i < num_packets; i++) { |
| 186 | av_add_index_entry(s->streams[0], pos, st->duration, 0, 0, AVINDEX_KEYFRAME); |
| 187 | pos += caf->bytes_per_packet ? caf->bytes_per_packet : ff_mp4_read_descr_len(pb); |
| 188 | st->duration += caf->frames_per_packet ? caf->frames_per_packet : ff_mp4_read_descr_len(pb); |
| 189 | } |
| 190 | |
| 191 | if (avio_tell(pb) - ccount > size) { |
| 192 | av_log(s, AV_LOG_ERROR, "error reading packet table\n"); |
| 193 | return AVERROR_INVALIDDATA; |
| 194 | } |
| 195 | avio_skip(pb, ccount + size - avio_tell(pb)); |
| 196 | |
| 197 | caf->num_bytes = pos; |
| 198 | return 0; |
| 199 | } |
| 200 | |
| 201 | /** Read information chunk */ |
| 202 | static void read_info_chunk(AVFormatContext *s, int64_t size) |
| 203 | { |
| 204 | AVIOContext *pb = s->pb; |
| 205 | unsigned int i; |
| 206 | unsigned int nb_entries = avio_rb32(pb); |
| 207 | for (i = 0; i < nb_entries && !avio_feof(pb); i++) { |
| 208 | char key[32]; |
| 209 | char value[1024]; |
| 210 | avio_get_str(pb, INT_MAX, key, sizeof(key)); |
| 211 | avio_get_str(pb, INT_MAX, value, sizeof(value)); |
| 212 | av_dict_set(&s->metadata, key, value, 0); |
| 213 | } |
| 214 | } |
| 215 | |
| 216 | static int read_header(AVFormatContext *s) |
| 217 | { |
| 218 | AVIOContext *pb = s->pb; |
| 219 | CafContext *caf = s->priv_data; |
| 220 | AVStream *st; |
| 221 | uint32_t tag = 0; |
| 222 | int found_data, ret; |
| 223 | int64_t size, pos; |
| 224 | |
| 225 | avio_skip(pb, 8); /* magic, version, file flags */ |
| 226 | |
| 227 | /* audio description chunk */ |
| 228 | if (avio_rb32(pb) != MKBETAG('d','e','s','c')) { |
| 229 | av_log(s, AV_LOG_ERROR, "desc chunk not present\n"); |
| 230 | return AVERROR_INVALIDDATA; |
| 231 | } |
| 232 | size = avio_rb64(pb); |
| 233 | if (size != 32) |
| 234 | return AVERROR_INVALIDDATA; |
| 235 | |
| 236 | ret = read_desc_chunk(s); |
| 237 | if (ret) |
| 238 | return ret; |
| 239 | st = s->streams[0]; |
| 240 | |
| 241 | /* parse each chunk */ |
| 242 | found_data = 0; |
| 243 | while (!avio_feof(pb)) { |
| 244 | |
| 245 | /* stop at data chunk if seeking is not supported or |
| 246 | data chunk size is unknown */ |
| 247 | if (found_data && (caf->data_size < 0 || !pb->seekable)) |
| 248 | break; |
| 249 | |
| 250 | tag = avio_rb32(pb); |
| 251 | size = avio_rb64(pb); |
| 252 | pos = avio_tell(pb); |
| 253 | if (avio_feof(pb)) |
| 254 | break; |
| 255 | |
| 256 | switch (tag) { |
| 257 | case MKBETAG('d','a','t','a'): |
| 258 | avio_skip(pb, 4); /* edit count */ |
| 259 | caf->data_start = avio_tell(pb); |
| 260 | caf->data_size = size < 0 ? -1 : size - 4; |
| 261 | if (caf->data_size > 0 && pb->seekable) |
| 262 | avio_skip(pb, caf->data_size); |
| 263 | found_data = 1; |
| 264 | break; |
| 265 | |
| 266 | case MKBETAG('c','h','a','n'): |
| 267 | if ((ret = ff_mov_read_chan(s, s->pb, st, size)) < 0) |
| 268 | return ret; |
| 269 | break; |
| 270 | |
| 271 | /* magic cookie chunk */ |
| 272 | case MKBETAG('k','u','k','i'): |
| 273 | if (read_kuki_chunk(s, size)) |
| 274 | return AVERROR_INVALIDDATA; |
| 275 | break; |
| 276 | |
| 277 | /* packet table chunk */ |
| 278 | case MKBETAG('p','a','k','t'): |
| 279 | if (read_pakt_chunk(s, size)) |
| 280 | return AVERROR_INVALIDDATA; |
| 281 | break; |
| 282 | |
| 283 | case MKBETAG('i','n','f','o'): |
| 284 | read_info_chunk(s, size); |
| 285 | break; |
| 286 | |
| 287 | default: |
| 288 | #define _(x) ((x) >= ' ' ? (x) : ' ') |
| 289 | av_log(s, AV_LOG_WARNING, |
| 290 | "skipping CAF chunk: %08"PRIX32" (%c%c%c%c), size %"PRId64"\n", |
| 291 | tag, _(tag>>24), _((tag>>16)&0xFF), _((tag>>8)&0xFF), _(tag&0xFF), size); |
| 292 | #undef _ |
| 293 | case MKBETAG('f','r','e','e'): |
| 294 | if (size < 0) |
| 295 | return AVERROR_INVALIDDATA; |
| 296 | break; |
| 297 | } |
| 298 | |
| 299 | if (size > 0) { |
| 300 | if (pos > INT64_MAX - size) |
| 301 | return AVERROR_INVALIDDATA; |
| 302 | avio_skip(pb, FFMAX(0, pos + size - avio_tell(pb))); |
| 303 | } |
| 304 | } |
| 305 | |
| 306 | if (!found_data) |
| 307 | return AVERROR_INVALIDDATA; |
| 308 | |
| 309 | if (caf->bytes_per_packet > 0 && caf->frames_per_packet > 0) { |
| 310 | if (caf->data_size > 0) |
| 311 | st->nb_frames = (caf->data_size / caf->bytes_per_packet) * caf->frames_per_packet; |
| 312 | } else if (st->nb_index_entries && st->duration > 0) { |
| 313 | st->codec->bit_rate = st->codec->sample_rate * caf->data_size * 8 / |
| 314 | st->duration; |
| 315 | } else { |
| 316 | av_log(s, AV_LOG_ERROR, "Missing packet table. It is required when " |
| 317 | "block size or frame size are variable.\n"); |
| 318 | return AVERROR_INVALIDDATA; |
| 319 | } |
| 320 | |
| 321 | avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate); |
| 322 | st->start_time = 0; |
| 323 | |
| 324 | /* position the stream at the start of data */ |
| 325 | if (caf->data_size >= 0) |
| 326 | avio_seek(pb, caf->data_start, SEEK_SET); |
| 327 | |
| 328 | return 0; |
| 329 | } |
| 330 | |
| 331 | #define CAF_MAX_PKT_SIZE 4096 |
| 332 | |
| 333 | static int read_packet(AVFormatContext *s, AVPacket *pkt) |
| 334 | { |
| 335 | AVIOContext *pb = s->pb; |
| 336 | AVStream *st = s->streams[0]; |
| 337 | CafContext *caf = s->priv_data; |
| 338 | int res, pkt_size = 0, pkt_frames = 0; |
| 339 | int64_t left = CAF_MAX_PKT_SIZE; |
| 340 | |
| 341 | if (avio_feof(pb)) |
| 342 | return AVERROR_EOF; |
| 343 | |
| 344 | /* don't read past end of data chunk */ |
| 345 | if (caf->data_size > 0) { |
| 346 | left = (caf->data_start + caf->data_size) - avio_tell(pb); |
| 347 | if (!left) |
| 348 | return AVERROR_EOF; |
| 349 | if (left < 0) |
| 350 | return AVERROR(EIO); |
| 351 | } |
| 352 | |
| 353 | pkt_frames = caf->frames_per_packet; |
| 354 | pkt_size = caf->bytes_per_packet; |
| 355 | |
| 356 | if (pkt_size > 0 && pkt_frames == 1) { |
| 357 | pkt_size = (CAF_MAX_PKT_SIZE / pkt_size) * pkt_size; |
| 358 | pkt_size = FFMIN(pkt_size, left); |
| 359 | pkt_frames = pkt_size / caf->bytes_per_packet; |
| 360 | } else if (st->nb_index_entries) { |
| 361 | if (caf->packet_cnt < st->nb_index_entries - 1) { |
| 362 | pkt_size = st->index_entries[caf->packet_cnt + 1].pos - st->index_entries[caf->packet_cnt].pos; |
| 363 | pkt_frames = st->index_entries[caf->packet_cnt + 1].timestamp - st->index_entries[caf->packet_cnt].timestamp; |
| 364 | } else if (caf->packet_cnt == st->nb_index_entries - 1) { |
| 365 | pkt_size = caf->num_bytes - st->index_entries[caf->packet_cnt].pos; |
| 366 | pkt_frames = st->duration - st->index_entries[caf->packet_cnt].timestamp; |
| 367 | } else { |
| 368 | return AVERROR(EIO); |
| 369 | } |
| 370 | } |
| 371 | |
| 372 | if (pkt_size == 0 || pkt_frames == 0 || pkt_size > left) |
| 373 | return AVERROR(EIO); |
| 374 | |
| 375 | res = av_get_packet(pb, pkt, pkt_size); |
| 376 | if (res < 0) |
| 377 | return res; |
| 378 | |
| 379 | pkt->size = res; |
| 380 | pkt->stream_index = 0; |
| 381 | pkt->dts = pkt->pts = caf->frame_cnt; |
| 382 | |
| 383 | caf->packet_cnt++; |
| 384 | caf->frame_cnt += pkt_frames; |
| 385 | |
| 386 | return 0; |
| 387 | } |
| 388 | |
| 389 | static int read_seek(AVFormatContext *s, int stream_index, |
| 390 | int64_t timestamp, int flags) |
| 391 | { |
| 392 | AVStream *st = s->streams[0]; |
| 393 | CafContext *caf = s->priv_data; |
| 394 | int64_t pos, packet_cnt, frame_cnt; |
| 395 | |
| 396 | timestamp = FFMAX(timestamp, 0); |
| 397 | |
| 398 | if (caf->frames_per_packet > 0 && caf->bytes_per_packet > 0) { |
| 399 | /* calculate new byte position based on target frame position */ |
| 400 | pos = caf->bytes_per_packet * (timestamp / caf->frames_per_packet); |
| 401 | if (caf->data_size > 0) |
| 402 | pos = FFMIN(pos, caf->data_size); |
| 403 | packet_cnt = pos / caf->bytes_per_packet; |
| 404 | frame_cnt = caf->frames_per_packet * packet_cnt; |
| 405 | } else if (st->nb_index_entries) { |
| 406 | packet_cnt = av_index_search_timestamp(st, timestamp, flags); |
| 407 | frame_cnt = st->index_entries[packet_cnt].timestamp; |
| 408 | pos = st->index_entries[packet_cnt].pos; |
| 409 | } else { |
| 410 | return -1; |
| 411 | } |
| 412 | |
| 413 | if (avio_seek(s->pb, pos + caf->data_start, SEEK_SET) < 0) |
| 414 | return -1; |
| 415 | |
| 416 | caf->packet_cnt = packet_cnt; |
| 417 | caf->frame_cnt = frame_cnt; |
| 418 | |
| 419 | return 0; |
| 420 | } |
| 421 | |
| 422 | AVInputFormat ff_caf_demuxer = { |
| 423 | .name = "caf", |
| 424 | .long_name = NULL_IF_CONFIG_SMALL("Apple CAF (Core Audio Format)"), |
| 425 | .priv_data_size = sizeof(CafContext), |
| 426 | .read_probe = probe, |
| 427 | .read_header = read_header, |
| 428 | .read_packet = read_packet, |
| 429 | .read_seek = read_seek, |
| 430 | .codec_tag = (const AVCodecTag* const []){ ff_codec_caf_tags, 0 }, |
| 431 | }; |