blob: 5847e88e11d3e01bc5f860615564b128686327e3
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 | #include "libavutil/ffmath.h" |
31 | |
32 | #include "opus.h" |
33 | #include "vorbis.h" |
34 | |
35 | static const uint16_t opus_frame_duration[32] = { |
36 | 480, 960, 1920, 2880, |
37 | 480, 960, 1920, 2880, |
38 | 480, 960, 1920, 2880, |
39 | 480, 960, |
40 | 480, 960, |
41 | 120, 240, 480, 960, |
42 | 120, 240, 480, 960, |
43 | 120, 240, 480, 960, |
44 | 120, 240, 480, 960, |
45 | }; |
46 | |
47 | /** |
48 | * Read a 1- or 2-byte frame length |
49 | */ |
50 | static inline int xiph_lacing_16bit(const uint8_t **ptr, const uint8_t *end) |
51 | { |
52 | int val; |
53 | |
54 | if (*ptr >= end) |
55 | return AVERROR_INVALIDDATA; |
56 | val = *(*ptr)++; |
57 | if (val >= 252) { |
58 | if (*ptr >= end) |
59 | return AVERROR_INVALIDDATA; |
60 | val += 4 * *(*ptr)++; |
61 | } |
62 | return val; |
63 | } |
64 | |
65 | /** |
66 | * Read a multi-byte length (used for code 3 packet padding size) |
67 | */ |
68 | static inline int xiph_lacing_full(const uint8_t **ptr, const uint8_t *end) |
69 | { |
70 | int val = 0; |
71 | int next; |
72 | |
73 | while (1) { |
74 | if (*ptr >= end || val > INT_MAX - 254) |
75 | return AVERROR_INVALIDDATA; |
76 | next = *(*ptr)++; |
77 | val += next; |
78 | if (next < 255) |
79 | break; |
80 | else |
81 | val--; |
82 | } |
83 | return val; |
84 | } |
85 | |
86 | /** |
87 | * Parse Opus packet info from raw packet data |
88 | */ |
89 | int ff_opus_parse_packet(OpusPacket *pkt, const uint8_t *buf, int buf_size, |
90 | int self_delimiting) |
91 | { |
92 | const uint8_t *ptr = buf; |
93 | const uint8_t *end = buf + buf_size; |
94 | int padding = 0; |
95 | int frame_bytes, i; |
96 | |
97 | if (buf_size < 1) |
98 | goto fail; |
99 | |
100 | /* TOC byte */ |
101 | i = *ptr++; |
102 | pkt->code = (i ) & 0x3; |
103 | pkt->stereo = (i >> 2) & 0x1; |
104 | pkt->config = (i >> 3) & 0x1F; |
105 | |
106 | /* code 2 and code 3 packets have at least 1 byte after the TOC */ |
107 | if (pkt->code >= 2 && buf_size < 2) |
108 | goto fail; |
109 | |
110 | switch (pkt->code) { |
111 | case 0: |
112 | /* 1 frame */ |
113 | pkt->frame_count = 1; |
114 | pkt->vbr = 0; |
115 | |
116 | if (self_delimiting) { |
117 | int len = xiph_lacing_16bit(&ptr, end); |
118 | if (len < 0 || len > end - ptr) |
119 | goto fail; |
120 | end = ptr + len; |
121 | buf_size = end - buf; |
122 | } |
123 | |
124 | frame_bytes = end - ptr; |
125 | if (frame_bytes > MAX_FRAME_SIZE) |
126 | goto fail; |
127 | pkt->frame_offset[0] = ptr - buf; |
128 | pkt->frame_size[0] = frame_bytes; |
129 | break; |
130 | case 1: |
131 | /* 2 frames, equal size */ |
132 | pkt->frame_count = 2; |
133 | pkt->vbr = 0; |
134 | |
135 | if (self_delimiting) { |
136 | int len = xiph_lacing_16bit(&ptr, end); |
137 | if (len < 0 || 2 * len > end - ptr) |
138 | goto fail; |
139 | end = ptr + 2 * len; |
140 | buf_size = end - buf; |
141 | } |
142 | |
143 | frame_bytes = end - ptr; |
144 | if (frame_bytes & 1 || frame_bytes >> 1 > MAX_FRAME_SIZE) |
145 | goto fail; |
146 | pkt->frame_offset[0] = ptr - buf; |
147 | pkt->frame_size[0] = frame_bytes >> 1; |
148 | pkt->frame_offset[1] = pkt->frame_offset[0] + pkt->frame_size[0]; |
149 | pkt->frame_size[1] = frame_bytes >> 1; |
150 | break; |
151 | case 2: |
152 | /* 2 frames, different sizes */ |
153 | pkt->frame_count = 2; |
154 | pkt->vbr = 1; |
155 | |
156 | /* read 1st frame size */ |
157 | frame_bytes = xiph_lacing_16bit(&ptr, end); |
158 | if (frame_bytes < 0) |
159 | goto fail; |
160 | |
161 | if (self_delimiting) { |
162 | int len = xiph_lacing_16bit(&ptr, end); |
163 | if (len < 0 || len + frame_bytes > end - ptr) |
164 | goto fail; |
165 | end = ptr + frame_bytes + len; |
166 | buf_size = end - buf; |
167 | } |
168 | |
169 | pkt->frame_offset[0] = ptr - buf; |
170 | pkt->frame_size[0] = frame_bytes; |
171 | |
172 | /* calculate 2nd frame size */ |
173 | frame_bytes = end - ptr - pkt->frame_size[0]; |
174 | if (frame_bytes < 0 || frame_bytes > MAX_FRAME_SIZE) |
175 | goto fail; |
176 | pkt->frame_offset[1] = pkt->frame_offset[0] + pkt->frame_size[0]; |
177 | pkt->frame_size[1] = frame_bytes; |
178 | break; |
179 | case 3: |
180 | /* 1 to 48 frames, can be different sizes */ |
181 | i = *ptr++; |
182 | pkt->frame_count = (i ) & 0x3F; |
183 | padding = (i >> 6) & 0x01; |
184 | pkt->vbr = (i >> 7) & 0x01; |
185 | |
186 | if (pkt->frame_count == 0 || pkt->frame_count > MAX_FRAMES) |
187 | goto fail; |
188 | |
189 | /* read padding size */ |
190 | if (padding) { |
191 | padding = xiph_lacing_full(&ptr, end); |
192 | if (padding < 0) |
193 | goto fail; |
194 | } |
195 | |
196 | /* read frame sizes */ |
197 | if (pkt->vbr) { |
198 | /* for VBR, all frames except the final one have their size coded |
199 | in the bitstream. the last frame size is implicit. */ |
200 | int total_bytes = 0; |
201 | for (i = 0; i < pkt->frame_count - 1; i++) { |
202 | frame_bytes = xiph_lacing_16bit(&ptr, end); |
203 | if (frame_bytes < 0) |
204 | goto fail; |
205 | pkt->frame_size[i] = frame_bytes; |
206 | total_bytes += frame_bytes; |
207 | } |
208 | |
209 | if (self_delimiting) { |
210 | int len = xiph_lacing_16bit(&ptr, end); |
211 | if (len < 0 || len + total_bytes + padding > end - ptr) |
212 | goto fail; |
213 | end = ptr + total_bytes + len + padding; |
214 | buf_size = end - buf; |
215 | } |
216 | |
217 | frame_bytes = end - ptr - padding; |
218 | if (total_bytes > frame_bytes) |
219 | goto fail; |
220 | pkt->frame_offset[0] = ptr - buf; |
221 | for (i = 1; i < pkt->frame_count; i++) |
222 | pkt->frame_offset[i] = pkt->frame_offset[i-1] + pkt->frame_size[i-1]; |
223 | pkt->frame_size[pkt->frame_count-1] = frame_bytes - total_bytes; |
224 | } else { |
225 | /* for CBR, the remaining packet bytes are divided evenly between |
226 | the frames */ |
227 | if (self_delimiting) { |
228 | frame_bytes = xiph_lacing_16bit(&ptr, end); |
229 | if (frame_bytes < 0 || pkt->frame_count * frame_bytes + padding > end - ptr) |
230 | goto fail; |
231 | end = ptr + pkt->frame_count * frame_bytes + padding; |
232 | buf_size = end - buf; |
233 | } else { |
234 | frame_bytes = end - ptr - padding; |
235 | if (frame_bytes % pkt->frame_count || |
236 | frame_bytes / pkt->frame_count > MAX_FRAME_SIZE) |
237 | goto fail; |
238 | frame_bytes /= pkt->frame_count; |
239 | } |
240 | |
241 | pkt->frame_offset[0] = ptr - buf; |
242 | pkt->frame_size[0] = frame_bytes; |
243 | for (i = 1; i < pkt->frame_count; i++) { |
244 | pkt->frame_offset[i] = pkt->frame_offset[i-1] + pkt->frame_size[i-1]; |
245 | pkt->frame_size[i] = frame_bytes; |
246 | } |
247 | } |
248 | } |
249 | |
250 | pkt->packet_size = buf_size; |
251 | pkt->data_size = pkt->packet_size - padding; |
252 | |
253 | /* total packet duration cannot be larger than 120ms */ |
254 | pkt->frame_duration = opus_frame_duration[pkt->config]; |
255 | if (pkt->frame_duration * pkt->frame_count > MAX_PACKET_DUR) |
256 | goto fail; |
257 | |
258 | /* set mode and bandwidth */ |
259 | if (pkt->config < 12) { |
260 | pkt->mode = OPUS_MODE_SILK; |
261 | pkt->bandwidth = pkt->config >> 2; |
262 | } else if (pkt->config < 16) { |
263 | pkt->mode = OPUS_MODE_HYBRID; |
264 | pkt->bandwidth = OPUS_BANDWIDTH_SUPERWIDEBAND + (pkt->config >= 14); |
265 | } else { |
266 | pkt->mode = OPUS_MODE_CELT; |
267 | pkt->bandwidth = (pkt->config - 16) >> 2; |
268 | /* skip medium band */ |
269 | if (pkt->bandwidth) |
270 | pkt->bandwidth++; |
271 | } |
272 | |
273 | return 0; |
274 | |
275 | fail: |
276 | memset(pkt, 0, sizeof(*pkt)); |
277 | return AVERROR_INVALIDDATA; |
278 | } |
279 | |
280 | static int channel_reorder_vorbis(int nb_channels, int channel_idx) |
281 | { |
282 | return ff_vorbis_channel_layout_offsets[nb_channels - 1][channel_idx]; |
283 | } |
284 | |
285 | static int channel_reorder_unknown(int nb_channels, int channel_idx) |
286 | { |
287 | return channel_idx; |
288 | } |
289 | |
290 | av_cold int ff_opus_parse_extradata(AVCodecContext *avctx, |
291 | OpusContext *s) |
292 | { |
293 | static const uint8_t default_channel_map[2] = { 0, 1 }; |
294 | |
295 | int (*channel_reorder)(int, int) = channel_reorder_unknown; |
296 | |
297 | const uint8_t *extradata, *channel_map; |
298 | int extradata_size; |
299 | int version, channels, map_type, streams, stereo_streams, i, j; |
300 | uint64_t layout; |
301 | |
302 | if (!avctx->extradata) { |
303 | if (avctx->channels > 2) { |
304 | av_log(avctx, AV_LOG_ERROR, |
305 | "Multichannel configuration without extradata.\n"); |
306 | return AVERROR(EINVAL); |
307 | } |
308 | extradata = opus_default_extradata; |
309 | extradata_size = sizeof(opus_default_extradata); |
310 | } else { |
311 | extradata = avctx->extradata; |
312 | extradata_size = avctx->extradata_size; |
313 | } |
314 | |
315 | if (extradata_size < 19) { |
316 | av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n", |
317 | extradata_size); |
318 | return AVERROR_INVALIDDATA; |
319 | } |
320 | |
321 | version = extradata[8]; |
322 | if (version > 15) { |
323 | avpriv_request_sample(avctx, "Extradata version %d", version); |
324 | return AVERROR_PATCHWELCOME; |
325 | } |
326 | |
327 | avctx->delay = AV_RL16(extradata + 10); |
328 | |
329 | channels = avctx->extradata ? extradata[9] : (avctx->channels == 1) ? 1 : 2; |
330 | if (!channels) { |
331 | av_log(avctx, AV_LOG_ERROR, "Zero channel count specified in the extradata\n"); |
332 | return AVERROR_INVALIDDATA; |
333 | } |
334 | |
335 | s->gain_i = AV_RL16(extradata + 16); |
336 | if (s->gain_i) |
337 | s->gain = ff_exp10(s->gain_i / (20.0 * 256)); |
338 | |
339 | map_type = extradata[18]; |
340 | if (!map_type) { |
341 | if (channels > 2) { |
342 | av_log(avctx, AV_LOG_ERROR, |
343 | "Channel mapping 0 is only specified for up to 2 channels\n"); |
344 | return AVERROR_INVALIDDATA; |
345 | } |
346 | layout = (channels == 1) ? AV_CH_LAYOUT_MONO : AV_CH_LAYOUT_STEREO; |
347 | streams = 1; |
348 | stereo_streams = channels - 1; |
349 | channel_map = default_channel_map; |
350 | } else if (map_type == 1 || map_type == 2 || map_type == 255) { |
351 | if (extradata_size < 21 + channels) { |
352 | av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n", |
353 | extradata_size); |
354 | return AVERROR_INVALIDDATA; |
355 | } |
356 | |
357 | streams = extradata[19]; |
358 | stereo_streams = extradata[20]; |
359 | if (!streams || stereo_streams > streams || |
360 | streams + stereo_streams > 255) { |
361 | av_log(avctx, AV_LOG_ERROR, |
362 | "Invalid stream/stereo stream count: %d/%d\n", streams, stereo_streams); |
363 | return AVERROR_INVALIDDATA; |
364 | } |
365 | |
366 | if (map_type == 1) { |
367 | if (channels > 8) { |
368 | av_log(avctx, AV_LOG_ERROR, |
369 | "Channel mapping 1 is only specified for up to 8 channels\n"); |
370 | return AVERROR_INVALIDDATA; |
371 | } |
372 | layout = ff_vorbis_channel_layouts[channels - 1]; |
373 | channel_reorder = channel_reorder_vorbis; |
374 | } else if (map_type == 2) { |
375 | int ambisonic_order = ff_sqrt(channels) - 1; |
376 | if (channels != ((ambisonic_order + 1) * (ambisonic_order + 1)) && |
377 | channels != ((ambisonic_order + 1) * (ambisonic_order + 1) + 2)) { |
378 | av_log(avctx, AV_LOG_ERROR, |
379 | "Channel mapping 2 is only specified for channel counts" |
380 | " which can be written as (n + 1)^2 or (n + 1)^2 + 2" |
381 | " for nonnegative integer n\n"); |
382 | return AVERROR_INVALIDDATA; |
383 | } |
384 | if (channels > 227) { |
385 | av_log(avctx, AV_LOG_ERROR, "Too many channels\n"); |
386 | return AVERROR_INVALIDDATA; |
387 | } |
388 | layout = 0; |
389 | } else |
390 | layout = 0; |
391 | |
392 | channel_map = extradata + 21; |
393 | } else { |
394 | avpriv_request_sample(avctx, "Mapping type %d", map_type); |
395 | return AVERROR_PATCHWELCOME; |
396 | } |
397 | |
398 | s->channel_maps = av_mallocz_array(channels, sizeof(*s->channel_maps)); |
399 | if (!s->channel_maps) |
400 | return AVERROR(ENOMEM); |
401 | |
402 | for (i = 0; i < channels; i++) { |
403 | ChannelMap *map = &s->channel_maps[i]; |
404 | uint8_t idx = channel_map[channel_reorder(channels, i)]; |
405 | |
406 | if (idx == 255) { |
407 | map->silence = 1; |
408 | continue; |
409 | } else if (idx >= streams + stereo_streams) { |
410 | av_log(avctx, AV_LOG_ERROR, |
411 | "Invalid channel map for output channel %d: %d\n", i, idx); |
412 | av_freep(&s->channel_maps); |
413 | return AVERROR_INVALIDDATA; |
414 | } |
415 | |
416 | /* check that we did not see this index yet */ |
417 | map->copy = 0; |
418 | for (j = 0; j < i; j++) |
419 | if (channel_map[channel_reorder(channels, j)] == idx) { |
420 | map->copy = 1; |
421 | map->copy_idx = j; |
422 | break; |
423 | } |
424 | |
425 | if (idx < 2 * stereo_streams) { |
426 | map->stream_idx = idx / 2; |
427 | map->channel_idx = idx & 1; |
428 | } else { |
429 | map->stream_idx = idx - stereo_streams; |
430 | map->channel_idx = 0; |
431 | } |
432 | } |
433 | |
434 | avctx->channels = channels; |
435 | avctx->channel_layout = layout; |
436 | s->nb_streams = streams; |
437 | s->nb_stereo_streams = stereo_streams; |
438 | |
439 | return 0; |
440 | } |
441 |