Line data Source code
1 : /*
2 : * TTA (The Lossless True Audio) decoder
3 : * Copyright (c) 2006 Alex Beregszaszi
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 : * TTA (The Lossless True Audio) decoder
25 : * @see http://www.true-audio.com/
26 : * @see http://tta.corecodec.org/
27 : * @author Alex Beregszaszi
28 : */
29 :
30 : #include <limits.h>
31 :
32 : #include "libavutil/crc.h"
33 : #include "libavutil/intreadwrite.h"
34 : #include "libavutil/opt.h"
35 :
36 : #define BITSTREAM_READER_LE
37 : #include "ttadata.h"
38 : #include "ttadsp.h"
39 : #include "avcodec.h"
40 : #include "get_bits.h"
41 : #include "thread.h"
42 : #include "unary.h"
43 : #include "internal.h"
44 :
45 : #define FORMAT_SIMPLE 1
46 : #define FORMAT_ENCRYPTED 2
47 :
48 : typedef struct TTAContext {
49 : AVClass *class;
50 : AVCodecContext *avctx;
51 : const AVCRC *crc_table;
52 :
53 : int format, channels, bps;
54 : unsigned data_length;
55 : int frame_length, last_frame_length;
56 :
57 : int32_t *decode_buffer;
58 :
59 : uint8_t crc_pass[8];
60 : uint8_t *pass;
61 : TTAChannel *ch_ctx;
62 : TTADSPContext dsp;
63 : } TTAContext;
64 :
65 : static const int64_t tta_channel_layouts[7] = {
66 : AV_CH_LAYOUT_STEREO,
67 : AV_CH_LAYOUT_STEREO|AV_CH_LOW_FREQUENCY,
68 : AV_CH_LAYOUT_QUAD,
69 : 0,
70 : AV_CH_LAYOUT_5POINT1_BACK,
71 : AV_CH_LAYOUT_5POINT1_BACK|AV_CH_BACK_CENTER,
72 : AV_CH_LAYOUT_7POINT1_WIDE
73 : };
74 :
75 0 : static int tta_check_crc(TTAContext *s, const uint8_t *buf, int buf_size)
76 : {
77 : uint32_t crc, CRC;
78 :
79 0 : CRC = AV_RL32(buf + buf_size);
80 0 : crc = av_crc(s->crc_table, 0xFFFFFFFFU, buf, buf_size);
81 0 : if (CRC != (crc ^ 0xFFFFFFFFU)) {
82 0 : av_log(s->avctx, AV_LOG_ERROR, "CRC error\n");
83 0 : return AVERROR_INVALIDDATA;
84 : }
85 :
86 0 : return 0;
87 : }
88 :
89 2 : static uint64_t tta_check_crc64(uint8_t *pass)
90 : {
91 2 : uint64_t crc = UINT64_MAX, poly = 0x42F0E1EBA9EA3693U;
92 2 : uint8_t *end = pass + strlen(pass);
93 : int i;
94 :
95 16 : while (pass < end) {
96 12 : crc ^= (uint64_t)*pass++ << 56;
97 108 : for (i = 0; i < 8; i++)
98 96 : crc = (crc << 1) ^ (poly & (((int64_t) crc) >> 63));
99 : }
100 :
101 2 : return crc ^ UINT64_MAX;
102 : }
103 :
104 12 : static int allocate_buffers(AVCodecContext *avctx)
105 : {
106 12 : TTAContext *s = avctx->priv_data;
107 :
108 12 : if (s->bps < 3) {
109 12 : s->decode_buffer = av_mallocz_array(sizeof(int32_t)*s->frame_length, s->channels);
110 12 : if (!s->decode_buffer)
111 0 : return AVERROR(ENOMEM);
112 : } else
113 0 : s->decode_buffer = NULL;
114 12 : s->ch_ctx = av_malloc_array(avctx->channels, sizeof(*s->ch_ctx));
115 12 : if (!s->ch_ctx) {
116 0 : av_freep(&s->decode_buffer);
117 0 : return AVERROR(ENOMEM);
118 : }
119 :
120 12 : return 0;
121 : }
122 :
123 12 : static av_cold int tta_decode_init(AVCodecContext * avctx)
124 : {
125 12 : TTAContext *s = avctx->priv_data;
126 : GetBitContext gb;
127 : int total_frames;
128 : int ret;
129 :
130 12 : s->avctx = avctx;
131 :
132 : // 30bytes includes TTA1 header
133 12 : if (avctx->extradata_size < 22)
134 0 : return AVERROR_INVALIDDATA;
135 :
136 12 : s->crc_table = av_crc_get_table(AV_CRC_32_IEEE_LE);
137 12 : ret = init_get_bits8(&gb, avctx->extradata, avctx->extradata_size);
138 12 : if (ret < 0)
139 0 : return ret;
140 :
141 12 : if (show_bits_long(&gb, 32) == AV_RL32("TTA1")) {
142 : /* signature */
143 12 : skip_bits_long(&gb, 32);
144 :
145 12 : s->format = get_bits(&gb, 16);
146 12 : if (s->format > 2) {
147 0 : av_log(avctx, AV_LOG_ERROR, "Invalid format\n");
148 0 : return AVERROR_INVALIDDATA;
149 : }
150 12 : if (s->format == FORMAT_ENCRYPTED) {
151 2 : if (!s->pass) {
152 0 : av_log(avctx, AV_LOG_ERROR, "Missing password for encrypted stream. Please use the -password option\n");
153 0 : return AVERROR(EINVAL);
154 : }
155 2 : AV_WL64(s->crc_pass, tta_check_crc64(s->pass));
156 : }
157 12 : avctx->channels = s->channels = get_bits(&gb, 16);
158 12 : if (s->channels > 1 && s->channels < 9)
159 8 : avctx->channel_layout = tta_channel_layouts[s->channels-2];
160 12 : avctx->bits_per_raw_sample = get_bits(&gb, 16);
161 12 : s->bps = (avctx->bits_per_raw_sample + 7) / 8;
162 12 : avctx->sample_rate = get_bits_long(&gb, 32);
163 12 : s->data_length = get_bits_long(&gb, 32);
164 12 : skip_bits_long(&gb, 32); // CRC32 of header
165 :
166 12 : if (s->channels == 0) {
167 0 : av_log(avctx, AV_LOG_ERROR, "Invalid number of channels\n");
168 0 : return AVERROR_INVALIDDATA;
169 12 : } else if (avctx->sample_rate == 0) {
170 0 : av_log(avctx, AV_LOG_ERROR, "Invalid samplerate\n");
171 0 : return AVERROR_INVALIDDATA;
172 : }
173 :
174 12 : switch(s->bps) {
175 0 : case 1: avctx->sample_fmt = AV_SAMPLE_FMT_U8; break;
176 12 : case 2:
177 12 : avctx->sample_fmt = AV_SAMPLE_FMT_S16;
178 12 : break;
179 0 : case 3:
180 0 : avctx->sample_fmt = AV_SAMPLE_FMT_S32;
181 0 : break;
182 : //case 4: avctx->sample_fmt = AV_SAMPLE_FMT_S32; break;
183 0 : default:
184 0 : av_log(avctx, AV_LOG_ERROR, "Invalid/unsupported sample format.\n");
185 0 : return AVERROR_INVALIDDATA;
186 : }
187 :
188 : // prevent overflow
189 12 : if (avctx->sample_rate > 0x7FFFFFu) {
190 0 : av_log(avctx, AV_LOG_ERROR, "sample_rate too large\n");
191 0 : return AVERROR(EINVAL);
192 : }
193 12 : s->frame_length = 256 * avctx->sample_rate / 245;
194 :
195 12 : s->last_frame_length = s->data_length % s->frame_length;
196 24 : total_frames = s->data_length / s->frame_length +
197 12 : (s->last_frame_length ? 1 : 0);
198 :
199 12 : av_log(avctx, AV_LOG_DEBUG, "format: %d chans: %d bps: %d rate: %d block: %d\n",
200 : s->format, avctx->channels, avctx->bits_per_coded_sample, avctx->sample_rate,
201 : avctx->block_align);
202 12 : av_log(avctx, AV_LOG_DEBUG, "data_length: %d frame_length: %d last: %d total: %d\n",
203 : s->data_length, s->frame_length, s->last_frame_length, total_frames);
204 :
205 12 : if(s->frame_length >= UINT_MAX / (s->channels * sizeof(int32_t))){
206 0 : av_log(avctx, AV_LOG_ERROR, "frame_length too large\n");
207 0 : return AVERROR_INVALIDDATA;
208 : }
209 : } else {
210 0 : av_log(avctx, AV_LOG_ERROR, "Wrong extradata present\n");
211 0 : return AVERROR_INVALIDDATA;
212 : }
213 :
214 12 : ff_ttadsp_init(&s->dsp);
215 :
216 12 : return allocate_buffers(avctx);
217 : }
218 :
219 40 : static int tta_decode_frame(AVCodecContext *avctx, void *data,
220 : int *got_frame_ptr, AVPacket *avpkt)
221 : {
222 40 : AVFrame *frame = data;
223 40 : ThreadFrame tframe = { .f = data };
224 40 : const uint8_t *buf = avpkt->data;
225 40 : int buf_size = avpkt->size;
226 40 : TTAContext *s = avctx->priv_data;
227 : GetBitContext gb;
228 : int i, ret;
229 40 : int cur_chan = 0, framelen = s->frame_length;
230 : int32_t *p;
231 :
232 40 : if (avctx->err_recognition & AV_EF_CRCCHECK) {
233 0 : if (buf_size < 4 ||
234 0 : (tta_check_crc(s, buf, buf_size - 4) && avctx->err_recognition & AV_EF_EXPLODE))
235 0 : return AVERROR_INVALIDDATA;
236 : }
237 :
238 40 : if ((ret = init_get_bits8(&gb, avpkt->data, avpkt->size)) < 0)
239 0 : return ret;
240 :
241 : /* get output buffer */
242 40 : frame->nb_samples = framelen;
243 40 : if ((ret = ff_thread_get_buffer(avctx, &tframe, 0)) < 0)
244 0 : return ret;
245 :
246 : // decode directly to output buffer for 24-bit sample format
247 40 : if (s->bps == 3)
248 0 : s->decode_buffer = (int32_t *)frame->data[0];
249 :
250 : // init per channel states
251 118 : for (i = 0; i < s->channels; i++) {
252 78 : TTAFilter *filter = &s->ch_ctx[i].filter;
253 78 : s->ch_ctx[i].predictor = 0;
254 78 : ff_tta_filter_init(filter, ff_tta_filter_configs[s->bps-1]);
255 78 : if (s->format == FORMAT_ENCRYPTED) {
256 : int i;
257 144 : for (i = 0; i < 8; i++)
258 128 : filter->qm[i] = sign_extend(s->crc_pass[i], 8);
259 : }
260 78 : ff_tta_rice_init(&s->ch_ctx[i].rice, 10, 10);
261 : }
262 :
263 40 : i = 0;
264 3367302 : for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++) {
265 3367268 : int32_t *predictor = &s->ch_ctx[cur_chan].predictor;
266 3367268 : TTAFilter *filter = &s->ch_ctx[cur_chan].filter;
267 3367268 : TTARice *rice = &s->ch_ctx[cur_chan].rice;
268 : uint32_t unary, depth, k;
269 : int32_t value;
270 :
271 3367268 : unary = get_unary(&gb, 0, get_bits_left(&gb));
272 :
273 3367268 : if (unary == 0) {
274 1554651 : depth = 0;
275 1554651 : k = rice->k0;
276 : } else {
277 1812617 : depth = 1;
278 1812617 : k = rice->k1;
279 1812617 : unary--;
280 : }
281 :
282 3367268 : if (get_bits_left(&gb) < k) {
283 0 : ret = AVERROR_INVALIDDATA;
284 0 : goto error;
285 : }
286 :
287 3367268 : if (k) {
288 3211598 : if (k > MIN_CACHE_BITS || unary > INT32_MAX >> k) {
289 0 : ret = AVERROR_INVALIDDATA;
290 0 : goto error;
291 : }
292 3211598 : value = (unary << k) + get_bits(&gb, k);
293 : } else
294 155670 : value = unary;
295 :
296 : // FIXME: copy paste from original
297 3367268 : switch (depth) {
298 1812617 : case 1:
299 1812617 : rice->sum1 += value - (rice->sum1 >> 4);
300 1812617 : if (rice->k1 > 0 && rice->sum1 < ff_tta_shift_16[rice->k1])
301 49342 : rice->k1--;
302 1763275 : else if(rice->sum1 > ff_tta_shift_16[rice->k1 + 1])
303 49047 : rice->k1++;
304 1812617 : value += ff_tta_shift_1[rice->k0];
305 3367268 : default:
306 3367268 : rice->sum0 += value - (rice->sum0 >> 4);
307 3367268 : if (rice->k0 > 0 && rice->sum0 < ff_tta_shift_16[rice->k0])
308 92975 : rice->k0--;
309 3274293 : else if(rice->sum0 > ff_tta_shift_16[rice->k0 + 1])
310 92683 : rice->k0++;
311 : }
312 :
313 : // extract coded value
314 3367268 : *p = 1 + ((value >> 1) ^ ((value & 1) - 1));
315 :
316 : // run hybrid filter
317 3367268 : s->dsp.filter_process(filter->qm, filter->dx, filter->dl, &filter->error, p,
318 : filter->shift, filter->round);
319 :
320 : // fixed order prediction
321 : #define PRED(x, k) (int32_t)((((uint64_t)(x) << (k)) - (x)) >> (k))
322 3367268 : switch (s->bps) {
323 0 : case 1: *p += PRED(*predictor, 4); break;
324 3367268 : case 2:
325 3367268 : case 3: *p += PRED(*predictor, 5); break;
326 0 : case 4: *p += *predictor; break;
327 : }
328 3367268 : *predictor = *p;
329 :
330 : // flip channels
331 3367268 : if (cur_chan < (s->channels-1))
332 1639534 : cur_chan++;
333 : else {
334 : // decorrelate in case of multiple channels
335 1727734 : if (s->channels > 1) {
336 1639534 : int32_t *r = p - 1;
337 3279068 : for (*p += *r / 2; r > p - s->channels; r--)
338 1639534 : *r = *(r + 1) - *r;
339 : }
340 1727734 : cur_chan = 0;
341 1727734 : i++;
342 : // check for last frame
343 1727734 : if (i == s->last_frame_length && get_bits_left(&gb) / 8 == 4) {
344 6 : frame->nb_samples = framelen = s->last_frame_length;
345 6 : break;
346 : }
347 : }
348 : }
349 :
350 40 : align_get_bits(&gb);
351 40 : if (get_bits_left(&gb) < 32) {
352 0 : ret = AVERROR_INVALIDDATA;
353 0 : goto error;
354 : }
355 40 : skip_bits_long(&gb, 32); // frame crc
356 :
357 : // convert to output buffer
358 40 : switch (s->bps) {
359 0 : case 1: {
360 0 : uint8_t *samples = (uint8_t *)frame->data[0];
361 0 : for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++)
362 0 : *samples++ = *p + 0x80;
363 0 : break;
364 : }
365 40 : case 2: {
366 40 : int16_t *samples = (int16_t *)frame->data[0];
367 3367308 : for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++)
368 3367268 : *samples++ = *p;
369 40 : break;
370 : }
371 0 : case 3: {
372 : // shift samples for 24-bit sample format
373 0 : int32_t *samples = (int32_t *)frame->data[0];
374 0 : for (i = 0; i < framelen * s->channels; i++)
375 0 : *samples++ <<= 8;
376 : // reset decode buffer
377 0 : s->decode_buffer = NULL;
378 0 : break;
379 : }
380 : }
381 :
382 40 : *got_frame_ptr = 1;
383 :
384 40 : return buf_size;
385 0 : error:
386 : // reset decode buffer
387 0 : if (s->bps == 3)
388 0 : s->decode_buffer = NULL;
389 0 : return ret;
390 : }
391 :
392 0 : static int init_thread_copy(AVCodecContext *avctx)
393 : {
394 0 : TTAContext *s = avctx->priv_data;
395 0 : s->avctx = avctx;
396 0 : return allocate_buffers(avctx);
397 : }
398 :
399 12 : static av_cold int tta_decode_close(AVCodecContext *avctx) {
400 12 : TTAContext *s = avctx->priv_data;
401 :
402 12 : if (s->bps < 3)
403 12 : av_freep(&s->decode_buffer);
404 12 : s->decode_buffer = NULL;
405 12 : av_freep(&s->ch_ctx);
406 :
407 12 : return 0;
408 : }
409 :
410 : #define OFFSET(x) offsetof(TTAContext, x)
411 : #define DEC (AV_OPT_FLAG_DECODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM)
412 : static const AVOption options[] = {
413 : { "password", "Set decoding password", OFFSET(pass), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, DEC },
414 : { NULL },
415 : };
416 :
417 : static const AVClass tta_decoder_class = {
418 : .class_name = "TTA Decoder",
419 : .item_name = av_default_item_name,
420 : .option = options,
421 : .version = LIBAVUTIL_VERSION_INT,
422 : };
423 :
424 : AVCodec ff_tta_decoder = {
425 : .name = "tta",
426 : .long_name = NULL_IF_CONFIG_SMALL("TTA (True Audio)"),
427 : .type = AVMEDIA_TYPE_AUDIO,
428 : .id = AV_CODEC_ID_TTA,
429 : .priv_data_size = sizeof(TTAContext),
430 : .init = tta_decode_init,
431 : .close = tta_decode_close,
432 : .decode = tta_decode_frame,
433 : .init_thread_copy = ONLY_IF_THREADS_ENABLED(init_thread_copy),
434 : .capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS,
435 : .priv_class = &tta_decoder_class,
436 : };
|