| 1 | /* |
| 2 | * Copyright (c) 2012 Andrew D'Addesio |
| 3 | * Copyright (c) 2013-2014 Mozilla Corporation |
| 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 | * Opus decoder/parser shared code |
| 25 | */ |
| 26 | |
| 27 | #include <stdint.h> |
| 28 | |
| 29 | #include "libavutil/error.h" |
| 30 | |
| 31 | #include "opus.h" |
| 32 | #include "vorbis.h" |
| 33 | |
| 34 | static const uint16_t opus_frame_duration[32] = { |
| 35 | 480, 960, 1920, 2880, |
| 36 | 480, 960, 1920, 2880, |
| 37 | 480, 960, 1920, 2880, |
| 38 | 480, 960, |
| 39 | 480, 960, |
| 40 | 120, 240, 480, 960, |
| 41 | 120, 240, 480, 960, |
| 42 | 120, 240, 480, 960, |
| 43 | 120, 240, 480, 960, |
| 44 | }; |
| 45 | |
| 46 | /** |
| 47 | * Read a 1- or 2-byte frame length |
| 48 | */ |
| 49 | static inline int xiph_lacing_16bit(const uint8_t **ptr, const uint8_t *end) |
| 50 | { |
| 51 | int val; |
| 52 | |
| 53 | if (*ptr >= end) |
| 54 | return AVERROR_INVALIDDATA; |
| 55 | val = *(*ptr)++; |
| 56 | if (val >= 252) { |
| 57 | if (*ptr >= end) |
| 58 | return AVERROR_INVALIDDATA; |
| 59 | val += 4 * *(*ptr)++; |
| 60 | } |
| 61 | return val; |
| 62 | } |
| 63 | |
| 64 | /** |
| 65 | * Read a multi-byte length (used for code 3 packet padding size) |
| 66 | */ |
| 67 | static inline int xiph_lacing_full(const uint8_t **ptr, const uint8_t *end) |
| 68 | { |
| 69 | int val = 0; |
| 70 | int next; |
| 71 | |
| 72 | while (1) { |
| 73 | if (*ptr >= end || val > INT_MAX - 254) |
| 74 | return AVERROR_INVALIDDATA; |
| 75 | next = *(*ptr)++; |
| 76 | val += next; |
| 77 | if (next < 255) |
| 78 | break; |
| 79 | else |
| 80 | val--; |
| 81 | } |
| 82 | return val; |
| 83 | } |
| 84 | |
| 85 | /** |
| 86 | * Parse Opus packet info from raw packet data |
| 87 | */ |
| 88 | int ff_opus_parse_packet(OpusPacket *pkt, const uint8_t *buf, int buf_size, |
| 89 | int self_delimiting) |
| 90 | { |
| 91 | const uint8_t *ptr = buf; |
| 92 | const uint8_t *end = buf + buf_size; |
| 93 | int padding = 0; |
| 94 | int frame_bytes, i; |
| 95 | |
| 96 | if (buf_size < 1) |
| 97 | goto fail; |
| 98 | |
| 99 | /* TOC byte */ |
| 100 | i = *ptr++; |
| 101 | pkt->code = (i ) & 0x3; |
| 102 | pkt->stereo = (i >> 2) & 0x1; |
| 103 | pkt->config = (i >> 3) & 0x1F; |
| 104 | |
| 105 | /* code 2 and code 3 packets have at least 1 byte after the TOC */ |
| 106 | if (pkt->code >= 2 && buf_size < 2) |
| 107 | goto fail; |
| 108 | |
| 109 | switch (pkt->code) { |
| 110 | case 0: |
| 111 | /* 1 frame */ |
| 112 | pkt->frame_count = 1; |
| 113 | pkt->vbr = 0; |
| 114 | |
| 115 | if (self_delimiting) { |
| 116 | int len = xiph_lacing_16bit(&ptr, end); |
| 117 | if (len < 0 || len > end - ptr) |
| 118 | goto fail; |
| 119 | end = ptr + len; |
| 120 | buf_size = end - buf; |
| 121 | } |
| 122 | |
| 123 | frame_bytes = end - ptr; |
| 124 | if (frame_bytes > MAX_FRAME_SIZE) |
| 125 | goto fail; |
| 126 | pkt->frame_offset[0] = ptr - buf; |
| 127 | pkt->frame_size[0] = frame_bytes; |
| 128 | break; |
| 129 | case 1: |
| 130 | /* 2 frames, equal size */ |
| 131 | pkt->frame_count = 2; |
| 132 | pkt->vbr = 0; |
| 133 | |
| 134 | if (self_delimiting) { |
| 135 | int len = xiph_lacing_16bit(&ptr, end); |
| 136 | if (len < 0 || 2 * len > end - ptr) |
| 137 | goto fail; |
| 138 | end = ptr + 2 * len; |
| 139 | buf_size = end - buf; |
| 140 | } |
| 141 | |
| 142 | frame_bytes = end - ptr; |
| 143 | if (frame_bytes & 1 || frame_bytes >> 1 > MAX_FRAME_SIZE) |
| 144 | goto fail; |
| 145 | pkt->frame_offset[0] = ptr - buf; |
| 146 | pkt->frame_size[0] = frame_bytes >> 1; |
| 147 | pkt->frame_offset[1] = pkt->frame_offset[0] + pkt->frame_size[0]; |
| 148 | pkt->frame_size[1] = frame_bytes >> 1; |
| 149 | break; |
| 150 | case 2: |
| 151 | /* 2 frames, different sizes */ |
| 152 | pkt->frame_count = 2; |
| 153 | pkt->vbr = 1; |
| 154 | |
| 155 | /* read 1st frame size */ |
| 156 | frame_bytes = xiph_lacing_16bit(&ptr, end); |
| 157 | if (frame_bytes < 0) |
| 158 | goto fail; |
| 159 | |
| 160 | if (self_delimiting) { |
| 161 | int len = xiph_lacing_16bit(&ptr, end); |
| 162 | if (len < 0 || len + frame_bytes > end - ptr) |
| 163 | goto fail; |
| 164 | end = ptr + frame_bytes + len; |
| 165 | buf_size = end - buf; |
| 166 | } |
| 167 | |
| 168 | pkt->frame_offset[0] = ptr - buf; |
| 169 | pkt->frame_size[0] = frame_bytes; |
| 170 | |
| 171 | /* calculate 2nd frame size */ |
| 172 | frame_bytes = end - ptr - pkt->frame_size[0]; |
| 173 | if (frame_bytes < 0 || frame_bytes > MAX_FRAME_SIZE) |
| 174 | goto fail; |
| 175 | pkt->frame_offset[1] = pkt->frame_offset[0] + pkt->frame_size[0]; |
| 176 | pkt->frame_size[1] = frame_bytes; |
| 177 | break; |
| 178 | case 3: |
| 179 | /* 1 to 48 frames, can be different sizes */ |
| 180 | i = *ptr++; |
| 181 | pkt->frame_count = (i ) & 0x3F; |
| 182 | padding = (i >> 6) & 0x01; |
| 183 | pkt->vbr = (i >> 7) & 0x01; |
| 184 | |
| 185 | if (pkt->frame_count == 0 || pkt->frame_count > MAX_FRAMES) |
| 186 | goto fail; |
| 187 | |
| 188 | /* read padding size */ |
| 189 | if (padding) { |
| 190 | padding = xiph_lacing_full(&ptr, end); |
| 191 | if (padding < 0) |
| 192 | goto fail; |
| 193 | } |
| 194 | |
| 195 | /* read frame sizes */ |
| 196 | if (pkt->vbr) { |
| 197 | /* for VBR, all frames except the final one have their size coded |
| 198 | in the bitstream. the last frame size is implicit. */ |
| 199 | int total_bytes = 0; |
| 200 | for (i = 0; i < pkt->frame_count - 1; i++) { |
| 201 | frame_bytes = xiph_lacing_16bit(&ptr, end); |
| 202 | if (frame_bytes < 0) |
| 203 | goto fail; |
| 204 | pkt->frame_size[i] = frame_bytes; |
| 205 | total_bytes += frame_bytes; |
| 206 | } |
| 207 | |
| 208 | if (self_delimiting) { |
| 209 | int len = xiph_lacing_16bit(&ptr, end); |
| 210 | if (len < 0 || len + total_bytes + padding > end - ptr) |
| 211 | goto fail; |
| 212 | end = ptr + total_bytes + len + padding; |
| 213 | buf_size = end - buf; |
| 214 | } |
| 215 | |
| 216 | frame_bytes = end - ptr - padding; |
| 217 | if (total_bytes > frame_bytes) |
| 218 | goto fail; |
| 219 | pkt->frame_offset[0] = ptr - buf; |
| 220 | for (i = 1; i < pkt->frame_count; i++) |
| 221 | pkt->frame_offset[i] = pkt->frame_offset[i-1] + pkt->frame_size[i-1]; |
| 222 | pkt->frame_size[pkt->frame_count-1] = frame_bytes - total_bytes; |
| 223 | } else { |
| 224 | /* for CBR, the remaining packet bytes are divided evenly between |
| 225 | the frames */ |
| 226 | if (self_delimiting) { |
| 227 | frame_bytes = xiph_lacing_16bit(&ptr, end); |
| 228 | if (frame_bytes < 0 || pkt->frame_count * frame_bytes + padding > end - ptr) |
| 229 | goto fail; |
| 230 | end = ptr + pkt->frame_count * frame_bytes + padding; |
| 231 | buf_size = end - buf; |
| 232 | } else { |
| 233 | frame_bytes = end - ptr - padding; |
| 234 | if (frame_bytes % pkt->frame_count || |
| 235 | frame_bytes / pkt->frame_count > MAX_FRAME_SIZE) |
| 236 | goto fail; |
| 237 | frame_bytes /= pkt->frame_count; |
| 238 | } |
| 239 | |
| 240 | pkt->frame_offset[0] = ptr - buf; |
| 241 | pkt->frame_size[0] = frame_bytes; |
| 242 | for (i = 1; i < pkt->frame_count; i++) { |
| 243 | pkt->frame_offset[i] = pkt->frame_offset[i-1] + pkt->frame_size[i-1]; |
| 244 | pkt->frame_size[i] = frame_bytes; |
| 245 | } |
| 246 | } |
| 247 | } |
| 248 | |
| 249 | pkt->packet_size = buf_size; |
| 250 | pkt->data_size = pkt->packet_size - padding; |
| 251 | |
| 252 | /* total packet duration cannot be larger than 120ms */ |
| 253 | pkt->frame_duration = opus_frame_duration[pkt->config]; |
| 254 | if (pkt->frame_duration * pkt->frame_count > MAX_PACKET_DUR) |
| 255 | goto fail; |
| 256 | |
| 257 | /* set mode and bandwidth */ |
| 258 | if (pkt->config < 12) { |
| 259 | pkt->mode = OPUS_MODE_SILK; |
| 260 | pkt->bandwidth = pkt->config >> 2; |
| 261 | } else if (pkt->config < 16) { |
| 262 | pkt->mode = OPUS_MODE_HYBRID; |
| 263 | pkt->bandwidth = OPUS_BANDWIDTH_SUPERWIDEBAND + (pkt->config >= 14); |
| 264 | } else { |
| 265 | pkt->mode = OPUS_MODE_CELT; |
| 266 | pkt->bandwidth = (pkt->config - 16) >> 2; |
| 267 | /* skip mediumband */ |
| 268 | if (pkt->bandwidth) |
| 269 | pkt->bandwidth++; |
| 270 | } |
| 271 | |
| 272 | return 0; |
| 273 | |
| 274 | fail: |
| 275 | memset(pkt, 0, sizeof(*pkt)); |
| 276 | return AVERROR_INVALIDDATA; |
| 277 | } |
| 278 | |
| 279 | static int channel_reorder_vorbis(int nb_channels, int channel_idx) |
| 280 | { |
| 281 | return ff_vorbis_channel_layout_offsets[nb_channels - 1][channel_idx]; |
| 282 | } |
| 283 | |
| 284 | static int channel_reorder_unknown(int nb_channels, int channel_idx) |
| 285 | { |
| 286 | return channel_idx; |
| 287 | } |
| 288 | |
| 289 | av_cold int ff_opus_parse_extradata(AVCodecContext *avctx, |
| 290 | OpusContext *s) |
| 291 | { |
| 292 | static const uint8_t default_channel_map[2] = { 0, 1 }; |
| 293 | |
| 294 | int (*channel_reorder)(int, int) = channel_reorder_unknown; |
| 295 | |
| 296 | const uint8_t *extradata, *channel_map; |
| 297 | int extradata_size; |
| 298 | int version, channels, map_type, streams, stereo_streams, i, j; |
| 299 | uint64_t layout; |
| 300 | |
| 301 | if (!avctx->extradata) { |
| 302 | if (avctx->channels > 2) { |
| 303 | av_log(avctx, AV_LOG_ERROR, |
| 304 | "Multichannel configuration without extradata.\n"); |
| 305 | return AVERROR(EINVAL); |
| 306 | } |
| 307 | extradata = opus_default_extradata; |
| 308 | extradata_size = sizeof(opus_default_extradata); |
| 309 | } else { |
| 310 | extradata = avctx->extradata; |
| 311 | extradata_size = avctx->extradata_size; |
| 312 | } |
| 313 | |
| 314 | if (extradata_size < 19) { |
| 315 | av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n", |
| 316 | extradata_size); |
| 317 | return AVERROR_INVALIDDATA; |
| 318 | } |
| 319 | |
| 320 | version = extradata[8]; |
| 321 | if (version > 15) { |
| 322 | avpriv_request_sample(avctx, "Extradata version %d", version); |
| 323 | return AVERROR_PATCHWELCOME; |
| 324 | } |
| 325 | |
| 326 | avctx->delay = AV_RL16(extradata + 10); |
| 327 | |
| 328 | channels = avctx->extradata ? extradata[9] : (avctx->channels == 1) ? 1 : 2; |
| 329 | if (!channels) { |
| 330 | av_log(avctx, AV_LOG_ERROR, "Zero channel count specified in the extadata\n"); |
| 331 | return AVERROR_INVALIDDATA; |
| 332 | } |
| 333 | |
| 334 | s->gain_i = AV_RL16(extradata + 16); |
| 335 | if (s->gain_i) |
| 336 | s->gain = pow(10, s->gain_i / (20.0 * 256)); |
| 337 | |
| 338 | map_type = extradata[18]; |
| 339 | if (!map_type) { |
| 340 | if (channels > 2) { |
| 341 | av_log(avctx, AV_LOG_ERROR, |
| 342 | "Channel mapping 0 is only specified for up to 2 channels\n"); |
| 343 | return AVERROR_INVALIDDATA; |
| 344 | } |
| 345 | layout = (channels == 1) ? AV_CH_LAYOUT_MONO : AV_CH_LAYOUT_STEREO; |
| 346 | streams = 1; |
| 347 | stereo_streams = channels - 1; |
| 348 | channel_map = default_channel_map; |
| 349 | } else if (map_type == 1 || map_type == 255) { |
| 350 | if (extradata_size < 21 + channels) { |
| 351 | av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n", |
| 352 | extradata_size); |
| 353 | return AVERROR_INVALIDDATA; |
| 354 | } |
| 355 | |
| 356 | streams = extradata[19]; |
| 357 | stereo_streams = extradata[20]; |
| 358 | if (!streams || stereo_streams > streams || |
| 359 | streams + stereo_streams > 255) { |
| 360 | av_log(avctx, AV_LOG_ERROR, |
| 361 | "Invalid stream/stereo stream count: %d/%d\n", streams, stereo_streams); |
| 362 | return AVERROR_INVALIDDATA; |
| 363 | } |
| 364 | |
| 365 | if (map_type == 1) { |
| 366 | if (channels > 8) { |
| 367 | av_log(avctx, AV_LOG_ERROR, |
| 368 | "Channel mapping 1 is only specified for up to 8 channels\n"); |
| 369 | return AVERROR_INVALIDDATA; |
| 370 | } |
| 371 | layout = ff_vorbis_channel_layouts[channels - 1]; |
| 372 | channel_reorder = channel_reorder_vorbis; |
| 373 | } else |
| 374 | layout = 0; |
| 375 | |
| 376 | channel_map = extradata + 21; |
| 377 | } else { |
| 378 | avpriv_request_sample(avctx, "Mapping type %d", map_type); |
| 379 | return AVERROR_PATCHWELCOME; |
| 380 | } |
| 381 | |
| 382 | s->channel_maps = av_mallocz_array(channels, sizeof(*s->channel_maps)); |
| 383 | if (!s->channel_maps) |
| 384 | return AVERROR(ENOMEM); |
| 385 | |
| 386 | for (i = 0; i < channels; i++) { |
| 387 | ChannelMap *map = &s->channel_maps[i]; |
| 388 | uint8_t idx = channel_map[channel_reorder(channels, i)]; |
| 389 | |
| 390 | if (idx == 255) { |
| 391 | map->silence = 1; |
| 392 | continue; |
| 393 | } else if (idx >= streams + stereo_streams) { |
| 394 | av_log(avctx, AV_LOG_ERROR, |
| 395 | "Invalid channel map for output channel %d: %d\n", i, idx); |
| 396 | return AVERROR_INVALIDDATA; |
| 397 | } |
| 398 | |
| 399 | /* check that we din't see this index yet */ |
| 400 | map->copy = 0; |
| 401 | for (j = 0; j < i; j++) |
| 402 | if (channel_map[channel_reorder(channels, j)] == idx) { |
| 403 | map->copy = 1; |
| 404 | map->copy_idx = j; |
| 405 | break; |
| 406 | } |
| 407 | |
| 408 | if (idx < 2 * stereo_streams) { |
| 409 | map->stream_idx = idx / 2; |
| 410 | map->channel_idx = idx & 1; |
| 411 | } else { |
| 412 | map->stream_idx = idx - stereo_streams; |
| 413 | map->channel_idx = 0; |
| 414 | } |
| 415 | } |
| 416 | |
| 417 | avctx->channels = channels; |
| 418 | avctx->channel_layout = layout; |
| 419 | s->nb_streams = streams; |
| 420 | s->nb_stereo_streams = stereo_streams; |
| 421 | |
| 422 | return 0; |
| 423 | } |