/src/ffmpeg/libavcodec/opus/parse.c
Line | Count | Source (jump to first uncovered line) |
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 "libavutil/attributes.h" |
28 | | #include "libavutil/channel_layout.h" |
29 | | #include "libavutil/error.h" |
30 | | #include "libavutil/intreadwrite.h" |
31 | | #include "libavutil/log.h" |
32 | | #include "libavutil/mem.h" |
33 | | |
34 | | #include "avcodec.h" |
35 | | #include "internal.h" |
36 | | #include "mathops.h" |
37 | | #include "opus.h" |
38 | | #include "parse.h" |
39 | | #include "vorbis_data.h" |
40 | | |
41 | | static const uint16_t opus_frame_duration[32] = { |
42 | | 480, 960, 1920, 2880, |
43 | | 480, 960, 1920, 2880, |
44 | | 480, 960, 1920, 2880, |
45 | | 480, 960, |
46 | | 480, 960, |
47 | | 120, 240, 480, 960, |
48 | | 120, 240, 480, 960, |
49 | | 120, 240, 480, 960, |
50 | | 120, 240, 480, 960, |
51 | | }; |
52 | | |
53 | | /** |
54 | | * Read a 1- or 2-byte frame length |
55 | | */ |
56 | | static inline int xiph_lacing_16bit(const uint8_t **ptr, const uint8_t *end) |
57 | 37.4k | { |
58 | 37.4k | int val; |
59 | | |
60 | 37.4k | if (*ptr >= end) |
61 | 572 | return AVERROR_INVALIDDATA; |
62 | 36.8k | val = *(*ptr)++; |
63 | 36.8k | if (val >= 252) { |
64 | 1.39k | if (*ptr >= end) |
65 | 139 | return AVERROR_INVALIDDATA; |
66 | 1.25k | val += 4 * *(*ptr)++; |
67 | 1.25k | } |
68 | 36.7k | return val; |
69 | 36.8k | } |
70 | | |
71 | | /** |
72 | | * Read a multi-byte length (used for code 3 packet padding size) |
73 | | */ |
74 | | static inline int xiph_lacing_full(const uint8_t **ptr, const uint8_t *end) |
75 | 2.03k | { |
76 | 2.03k | int val = 0; |
77 | 2.03k | int next; |
78 | | |
79 | 2.30k | while (1) { |
80 | 2.30k | if (*ptr >= end || val > INT_MAX - 254) |
81 | 93 | return AVERROR_INVALIDDATA; |
82 | 2.20k | next = *(*ptr)++; |
83 | 2.20k | val += next; |
84 | 2.20k | if (next < 255) |
85 | 1.94k | break; |
86 | 266 | else |
87 | 266 | val--; |
88 | 2.20k | } |
89 | 1.94k | return val; |
90 | 2.03k | } |
91 | | |
92 | | /** |
93 | | * Parse Opus packet info from raw packet data |
94 | | */ |
95 | | int ff_opus_parse_packet(OpusPacket *pkt, const uint8_t *buf, int buf_size, |
96 | | int self_delimiting) |
97 | 44.8k | { |
98 | 44.8k | const uint8_t *ptr = buf; |
99 | 44.8k | const uint8_t *end = buf + buf_size; |
100 | 44.8k | int padding = 0; |
101 | 44.8k | int frame_bytes, i; |
102 | | |
103 | 44.8k | if (buf_size < 1) |
104 | 333 | goto fail; |
105 | | |
106 | | /* TOC byte */ |
107 | 44.5k | i = *ptr++; |
108 | 44.5k | pkt->code = (i ) & 0x3; |
109 | 44.5k | pkt->stereo = (i >> 2) & 0x1; |
110 | 44.5k | pkt->config = (i >> 3) & 0x1F; |
111 | | |
112 | | /* code 2 and code 3 packets have at least 1 byte after the TOC */ |
113 | 44.5k | if (pkt->code >= 2 && buf_size < 2) |
114 | 661 | goto fail; |
115 | | |
116 | 43.8k | switch (pkt->code) { |
117 | 23.2k | case 0: |
118 | | /* 1 frame */ |
119 | 23.2k | pkt->frame_count = 1; |
120 | 23.2k | pkt->vbr = 0; |
121 | | |
122 | 23.2k | if (self_delimiting) { |
123 | 7.60k | int len = xiph_lacing_16bit(&ptr, end); |
124 | 7.60k | if (len < 0 || len > end - ptr) |
125 | 3.53k | goto fail; |
126 | 4.07k | end = ptr + len; |
127 | 4.07k | buf_size = end - buf; |
128 | 4.07k | } |
129 | | |
130 | 19.7k | frame_bytes = end - ptr; |
131 | 19.7k | if (frame_bytes > OPUS_MAX_FRAME_SIZE) |
132 | 95 | goto fail; |
133 | 19.6k | pkt->frame_offset[0] = ptr - buf; |
134 | 19.6k | pkt->frame_size[0] = frame_bytes; |
135 | 19.6k | break; |
136 | 6.21k | case 1: |
137 | | /* 2 frames, equal size */ |
138 | 6.21k | pkt->frame_count = 2; |
139 | 6.21k | pkt->vbr = 0; |
140 | | |
141 | 6.21k | if (self_delimiting) { |
142 | 1.07k | int len = xiph_lacing_16bit(&ptr, end); |
143 | 1.07k | if (len < 0 || 2 * len > end - ptr) |
144 | 1.06k | goto fail; |
145 | 9 | end = ptr + 2 * len; |
146 | 9 | buf_size = end - buf; |
147 | 9 | } |
148 | | |
149 | 5.14k | frame_bytes = end - ptr; |
150 | 5.14k | if (frame_bytes & 1 || frame_bytes >> 1 > OPUS_MAX_FRAME_SIZE) |
151 | 2.99k | goto fail; |
152 | 2.14k | pkt->frame_offset[0] = ptr - buf; |
153 | 2.14k | pkt->frame_size[0] = frame_bytes >> 1; |
154 | 2.14k | pkt->frame_offset[1] = pkt->frame_offset[0] + pkt->frame_size[0]; |
155 | 2.14k | pkt->frame_size[1] = frame_bytes >> 1; |
156 | 2.14k | break; |
157 | 6.04k | case 2: |
158 | | /* 2 frames, different sizes */ |
159 | 6.04k | pkt->frame_count = 2; |
160 | 6.04k | pkt->vbr = 1; |
161 | | |
162 | | /* read 1st frame size */ |
163 | 6.04k | frame_bytes = xiph_lacing_16bit(&ptr, end); |
164 | 6.04k | if (frame_bytes < 0) |
165 | 15 | goto fail; |
166 | | |
167 | 6.03k | if (self_delimiting) { |
168 | 1.49k | int len = xiph_lacing_16bit(&ptr, end); |
169 | 1.49k | if (len < 0 || len + frame_bytes > end - ptr) |
170 | 1.46k | goto fail; |
171 | 22 | end = ptr + frame_bytes + len; |
172 | 22 | buf_size = end - buf; |
173 | 22 | } |
174 | | |
175 | 4.56k | pkt->frame_offset[0] = ptr - buf; |
176 | 4.56k | pkt->frame_size[0] = frame_bytes; |
177 | | |
178 | | /* calculate 2nd frame size */ |
179 | 4.56k | frame_bytes = end - ptr - pkt->frame_size[0]; |
180 | 4.56k | if (frame_bytes < 0 || frame_bytes > OPUS_MAX_FRAME_SIZE) |
181 | 1.90k | goto fail; |
182 | 2.66k | pkt->frame_offset[1] = pkt->frame_offset[0] + pkt->frame_size[0]; |
183 | 2.66k | pkt->frame_size[1] = frame_bytes; |
184 | 2.66k | break; |
185 | 8.35k | case 3: |
186 | | /* 1 to 48 frames, can be different sizes */ |
187 | 8.35k | i = *ptr++; |
188 | 8.35k | pkt->frame_count = (i ) & 0x3F; |
189 | 8.35k | padding = (i >> 6) & 0x01; |
190 | 8.35k | pkt->vbr = (i >> 7) & 0x01; |
191 | | |
192 | 8.35k | if (pkt->frame_count == 0 || pkt->frame_count > OPUS_MAX_FRAMES) |
193 | 2.69k | goto fail; |
194 | | |
195 | | /* read padding size */ |
196 | 5.65k | if (padding) { |
197 | 2.03k | padding = xiph_lacing_full(&ptr, end); |
198 | 2.03k | if (padding < 0) |
199 | 93 | goto fail; |
200 | 2.03k | } |
201 | | |
202 | | /* read frame sizes */ |
203 | 5.56k | if (pkt->vbr) { |
204 | | /* for VBR, all frames except the final one have their size coded |
205 | | in the bitstream. the last frame size is implicit. */ |
206 | 2.41k | int total_bytes = 0; |
207 | 22.6k | for (i = 0; i < pkt->frame_count - 1; i++) { |
208 | 20.5k | frame_bytes = xiph_lacing_16bit(&ptr, end); |
209 | 20.5k | if (frame_bytes < 0) |
210 | 364 | goto fail; |
211 | 20.2k | pkt->frame_size[i] = frame_bytes; |
212 | 20.2k | total_bytes += frame_bytes; |
213 | 20.2k | } |
214 | | |
215 | 2.04k | if (self_delimiting) { |
216 | 34 | int len = xiph_lacing_16bit(&ptr, end); |
217 | 34 | if (len < 0 || len + total_bytes + padding > end - ptr) |
218 | 34 | goto fail; |
219 | 0 | end = ptr + total_bytes + len + padding; |
220 | 0 | buf_size = end - buf; |
221 | 0 | } |
222 | | |
223 | 2.01k | frame_bytes = end - ptr - padding; |
224 | 2.01k | if (total_bytes > frame_bytes) |
225 | 622 | goto fail; |
226 | 1.39k | pkt->frame_offset[0] = ptr - buf; |
227 | 5.60k | for (i = 1; i < pkt->frame_count; i++) |
228 | 4.21k | pkt->frame_offset[i] = pkt->frame_offset[i-1] + pkt->frame_size[i-1]; |
229 | 1.39k | pkt->frame_size[pkt->frame_count-1] = frame_bytes - total_bytes; |
230 | 3.15k | } else { |
231 | | /* for CBR, the remaining packet bytes are divided evenly between |
232 | | the frames */ |
233 | 3.15k | if (self_delimiting) { |
234 | 618 | frame_bytes = xiph_lacing_16bit(&ptr, end); |
235 | 618 | if (frame_bytes < 0 || pkt->frame_count * frame_bytes + padding > end - ptr) |
236 | 398 | goto fail; |
237 | 220 | end = ptr + pkt->frame_count * frame_bytes + padding; |
238 | 220 | buf_size = end - buf; |
239 | 2.53k | } else { |
240 | 2.53k | frame_bytes = end - ptr - padding; |
241 | 2.53k | if (frame_bytes % pkt->frame_count || |
242 | 2.53k | frame_bytes / pkt->frame_count > OPUS_MAX_FRAME_SIZE) |
243 | 1.27k | goto fail; |
244 | 1.26k | frame_bytes /= pkt->frame_count; |
245 | 1.26k | } |
246 | | |
247 | 1.48k | pkt->frame_offset[0] = ptr - buf; |
248 | 1.48k | pkt->frame_size[0] = frame_bytes; |
249 | 6.80k | for (i = 1; i < pkt->frame_count; i++) { |
250 | 5.32k | pkt->frame_offset[i] = pkt->frame_offset[i-1] + pkt->frame_size[i-1]; |
251 | 5.32k | pkt->frame_size[i] = frame_bytes; |
252 | 5.32k | } |
253 | 1.48k | } |
254 | 43.8k | } |
255 | | |
256 | 27.3k | pkt->packet_size = buf_size; |
257 | 27.3k | pkt->data_size = pkt->packet_size - padding; |
258 | | |
259 | | /* total packet duration cannot be larger than 120ms */ |
260 | 27.3k | pkt->frame_duration = opus_frame_duration[pkt->config]; |
261 | 27.3k | if (pkt->frame_duration * pkt->frame_count > OPUS_MAX_PACKET_DUR) |
262 | 291 | goto fail; |
263 | | |
264 | | /* set mode and bandwidth */ |
265 | 27.0k | if (pkt->config < 12) { |
266 | 14.4k | pkt->mode = OPUS_MODE_SILK; |
267 | 14.4k | pkt->bandwidth = pkt->config >> 2; |
268 | 14.4k | } else if (pkt->config < 16) { |
269 | 2.89k | pkt->mode = OPUS_MODE_HYBRID; |
270 | 2.89k | pkt->bandwidth = OPUS_BANDWIDTH_SUPERWIDEBAND + (pkt->config >= 14); |
271 | 9.65k | } else { |
272 | 9.65k | pkt->mode = OPUS_MODE_CELT; |
273 | 9.65k | pkt->bandwidth = (pkt->config - 16) >> 2; |
274 | | /* skip medium band */ |
275 | 9.65k | if (pkt->bandwidth) |
276 | 8.54k | pkt->bandwidth++; |
277 | 9.65k | } |
278 | | |
279 | 27.0k | return 0; |
280 | | |
281 | 17.8k | fail: |
282 | 17.8k | memset(pkt, 0, sizeof(*pkt)); |
283 | 17.8k | return AVERROR_INVALIDDATA; |
284 | 27.3k | } |
285 | | |
286 | | static int channel_reorder_vorbis(int nb_channels, int channel_idx) |
287 | 657 | { |
288 | 657 | return ff_vorbis_channel_layout_offsets[nb_channels - 1][channel_idx]; |
289 | 657 | } |
290 | | |
291 | | static int channel_reorder_unknown(int nb_channels, int channel_idx) |
292 | 25 | { |
293 | 25 | return channel_idx; |
294 | 25 | } |
295 | | |
296 | | av_cold int ff_opus_parse_extradata(AVCodecContext *avctx, |
297 | | OpusParseContext *s) |
298 | 153 | { |
299 | 153 | static const uint8_t default_channel_map[2] = { 0, 1 }; |
300 | | |
301 | 153 | int (*channel_reorder)(int, int) = channel_reorder_unknown; |
302 | 153 | int channels = avctx->ch_layout.nb_channels; |
303 | | |
304 | 153 | const uint8_t *extradata, *channel_map; |
305 | 153 | int extradata_size; |
306 | 153 | int version, map_type, streams, stereo_streams, i, j, ret; |
307 | 153 | AVChannelLayout layout = { 0 }; |
308 | | |
309 | 153 | if (!avctx->extradata) { |
310 | 0 | if (channels > 2) { |
311 | 0 | av_log(avctx, AV_LOG_ERROR, |
312 | 0 | "Multichannel configuration without extradata.\n"); |
313 | 0 | return AVERROR(EINVAL); |
314 | 0 | } |
315 | 0 | extradata = opus_default_extradata; |
316 | 0 | extradata_size = sizeof(opus_default_extradata); |
317 | 153 | } else { |
318 | 153 | extradata = avctx->extradata; |
319 | 153 | extradata_size = avctx->extradata_size; |
320 | 153 | } |
321 | | |
322 | 153 | if (extradata_size < 19) { |
323 | 0 | av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n", |
324 | 0 | extradata_size); |
325 | 0 | return AVERROR_INVALIDDATA; |
326 | 0 | } |
327 | | |
328 | 153 | version = extradata[8]; |
329 | 153 | if (version > 15) { |
330 | 93 | avpriv_request_sample(avctx, "Extradata version %d", version); |
331 | 93 | return AVERROR_PATCHWELCOME; |
332 | 93 | } |
333 | | |
334 | 60 | avctx->delay = AV_RL16(extradata + 10); |
335 | 60 | if (avctx->internal) |
336 | 0 | avctx->internal->skip_samples = avctx->delay; |
337 | | |
338 | 60 | channels = avctx->extradata ? extradata[9] : (channels == 1) ? 1 : 2; |
339 | 60 | if (!channels) { |
340 | 0 | av_log(avctx, AV_LOG_ERROR, "Zero channel count specified in the extradata\n"); |
341 | 0 | return AVERROR_INVALIDDATA; |
342 | 0 | } |
343 | | |
344 | 60 | s->gain_i = AV_RL16(extradata + 16); |
345 | | |
346 | 60 | map_type = extradata[18]; |
347 | 60 | if (!map_type) { |
348 | 13 | if (channels > 2) { |
349 | 0 | av_log(avctx, AV_LOG_ERROR, |
350 | 0 | "Channel mapping 0 is only specified for up to 2 channels\n"); |
351 | 0 | ret = AVERROR_INVALIDDATA; |
352 | 0 | goto fail; |
353 | 0 | } |
354 | 13 | layout = (channels == 1) ? (AVChannelLayout)AV_CHANNEL_LAYOUT_MONO : |
355 | 13 | (AVChannelLayout)AV_CHANNEL_LAYOUT_STEREO; |
356 | 13 | streams = 1; |
357 | 13 | stereo_streams = channels - 1; |
358 | 13 | channel_map = default_channel_map; |
359 | 47 | } else if (map_type == 1 || map_type == 2 || map_type == 255) { |
360 | 27 | if (extradata_size < 21 + channels) { |
361 | 5 | av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n", |
362 | 5 | extradata_size); |
363 | 5 | ret = AVERROR_INVALIDDATA; |
364 | 5 | goto fail; |
365 | 5 | } |
366 | | |
367 | 22 | streams = extradata[19]; |
368 | 22 | stereo_streams = extradata[20]; |
369 | 22 | if (!streams || stereo_streams > streams || |
370 | 22 | streams + stereo_streams > 255) { |
371 | 0 | av_log(avctx, AV_LOG_ERROR, |
372 | 0 | "Invalid stream/stereo stream count: %d/%d\n", streams, stereo_streams); |
373 | 0 | ret = AVERROR_INVALIDDATA; |
374 | 0 | goto fail; |
375 | 0 | } |
376 | | |
377 | 22 | if (map_type == 1) { |
378 | 22 | if (channels > 8) { |
379 | 0 | av_log(avctx, AV_LOG_ERROR, |
380 | 0 | "Channel mapping 1 is only specified for up to 8 channels\n"); |
381 | 0 | ret = AVERROR_INVALIDDATA; |
382 | 0 | goto fail; |
383 | 0 | } |
384 | 22 | av_channel_layout_copy(&layout, &ff_vorbis_ch_layouts[channels - 1]); |
385 | 22 | channel_reorder = channel_reorder_vorbis; |
386 | 22 | } else if (map_type == 2) { |
387 | 0 | int ambisonic_order = ff_sqrt(channels) - 1; |
388 | 0 | if (channels != ((ambisonic_order + 1) * (ambisonic_order + 1)) && |
389 | 0 | channels != ((ambisonic_order + 1) * (ambisonic_order + 1) + 2)) { |
390 | 0 | av_log(avctx, AV_LOG_ERROR, |
391 | 0 | "Channel mapping 2 is only specified for channel counts" |
392 | 0 | " which can be written as (n + 1)^2 or (n + 1)^2 + 2" |
393 | 0 | " for nonnegative integer n\n"); |
394 | 0 | ret = AVERROR_INVALIDDATA; |
395 | 0 | goto fail; |
396 | 0 | } |
397 | 0 | if (channels > 227) { |
398 | 0 | av_log(avctx, AV_LOG_ERROR, "Too many channels\n"); |
399 | 0 | ret = AVERROR_INVALIDDATA; |
400 | 0 | goto fail; |
401 | 0 | } |
402 | | |
403 | 0 | layout.order = AV_CHANNEL_ORDER_AMBISONIC; |
404 | 0 | layout.nb_channels = channels; |
405 | 0 | if (channels != ((ambisonic_order + 1) * (ambisonic_order + 1))) |
406 | 0 | layout.u.mask = AV_CH_LAYOUT_STEREO; |
407 | 0 | } else { |
408 | 0 | layout.order = AV_CHANNEL_ORDER_UNSPEC; |
409 | 0 | layout.nb_channels = channels; |
410 | 0 | } |
411 | | |
412 | 22 | channel_map = extradata + 21; |
413 | 22 | } else { |
414 | 20 | avpriv_request_sample(avctx, "Mapping type %d", map_type); |
415 | 20 | return AVERROR_PATCHWELCOME; |
416 | 20 | } |
417 | | |
418 | 35 | s->channel_maps = av_calloc(channels, sizeof(*s->channel_maps)); |
419 | 35 | if (!s->channel_maps) { |
420 | 0 | ret = AVERROR(ENOMEM); |
421 | 0 | goto fail; |
422 | 0 | } |
423 | | |
424 | 212 | for (i = 0; i < channels; i++) { |
425 | 177 | ChannelMap *map = &s->channel_maps[i]; |
426 | 177 | uint8_t idx = channel_map[channel_reorder(channels, i)]; |
427 | | |
428 | 177 | if (idx == 255) { |
429 | 12 | map->silence = 1; |
430 | 12 | continue; |
431 | 165 | } else if (idx >= streams + stereo_streams) { |
432 | 0 | av_log(avctx, AV_LOG_ERROR, |
433 | 0 | "Invalid channel map for output channel %d: %d\n", i, idx); |
434 | 0 | av_freep(&s->channel_maps); |
435 | 0 | ret = AVERROR_INVALIDDATA; |
436 | 0 | goto fail; |
437 | 0 | } |
438 | | |
439 | | /* check that we did not see this index yet */ |
440 | 165 | map->copy = 0; |
441 | 670 | for (j = 0; j < i; j++) |
442 | 505 | if (channel_map[channel_reorder(channels, j)] == idx) { |
443 | 0 | map->copy = 1; |
444 | 0 | map->copy_idx = j; |
445 | 0 | break; |
446 | 0 | } |
447 | | |
448 | 165 | if (idx < 2 * stereo_streams) { |
449 | 118 | map->stream_idx = idx / 2; |
450 | 118 | map->channel_idx = idx & 1; |
451 | 118 | } else { |
452 | 47 | map->stream_idx = idx - stereo_streams; |
453 | 47 | map->channel_idx = 0; |
454 | 47 | } |
455 | 165 | } |
456 | | |
457 | 35 | ret = av_channel_layout_copy(&avctx->ch_layout, &layout); |
458 | 35 | if (ret < 0) |
459 | 0 | goto fail; |
460 | | |
461 | 35 | s->nb_streams = streams; |
462 | 35 | s->nb_stereo_streams = stereo_streams; |
463 | | |
464 | 35 | return 0; |
465 | 5 | fail: |
466 | 5 | av_channel_layout_uninit(&layout); |
467 | 5 | return ret; |
468 | 35 | } |
469 | | |