xref: /third_party/ffmpeg/libavcodec/tta.c (revision cabdff1a)
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/channel_layout.h"
33#include "libavutil/crc.h"
34#include "libavutil/intreadwrite.h"
35#include "libavutil/opt.h"
36
37#define BITSTREAM_READER_LE
38#include "ttadata.h"
39#include "ttadsp.h"
40#include "avcodec.h"
41#include "codec_internal.h"
42#include "get_bits.h"
43#include "thread.h"
44#include "unary.h"
45
46#define FORMAT_SIMPLE    1
47#define FORMAT_ENCRYPTED 2
48
49typedef struct TTAContext {
50    AVClass *class;
51    AVCodecContext *avctx;
52    const AVCRC *crc_table;
53
54    int format, channels, bps;
55    unsigned data_length;
56    int frame_length, last_frame_length;
57
58    int32_t *decode_buffer;
59
60    uint8_t crc_pass[8];
61    uint8_t *pass;
62    TTAChannel *ch_ctx;
63    TTADSPContext dsp;
64} TTAContext;
65
66static const int64_t tta_channel_layouts[7] = {
67    AV_CH_LAYOUT_STEREO,
68    AV_CH_LAYOUT_STEREO|AV_CH_LOW_FREQUENCY,
69    AV_CH_LAYOUT_QUAD,
70    0,
71    AV_CH_LAYOUT_5POINT1_BACK,
72    AV_CH_LAYOUT_5POINT1_BACK|AV_CH_BACK_CENTER,
73    AV_CH_LAYOUT_7POINT1_WIDE
74};
75
76static int tta_check_crc(TTAContext *s, const uint8_t *buf, int buf_size)
77{
78    uint32_t crc, CRC;
79
80    CRC = AV_RL32(buf + buf_size);
81    crc = av_crc(s->crc_table, 0xFFFFFFFFU, buf, buf_size);
82    if (CRC != (crc ^ 0xFFFFFFFFU)) {
83        av_log(s->avctx, AV_LOG_ERROR, "CRC error\n");
84        return AVERROR_INVALIDDATA;
85    }
86
87    return 0;
88}
89
90static uint64_t tta_check_crc64(uint8_t *pass)
91{
92    uint64_t crc = UINT64_MAX, poly = 0x42F0E1EBA9EA3693U;
93    uint8_t *end = pass + strlen(pass);
94    int i;
95
96    while (pass < end) {
97        crc ^= (uint64_t)*pass++ << 56;
98        for (i = 0; i < 8; i++)
99            crc = (crc << 1) ^ (poly & (((int64_t) crc) >> 63));
100    }
101
102    return crc ^ UINT64_MAX;
103}
104
105static int allocate_buffers(AVCodecContext *avctx)
106{
107    TTAContext *s = avctx->priv_data;
108
109    if (s->bps < 3) {
110        s->decode_buffer = av_calloc(s->frame_length,
111                                     sizeof(*s->decode_buffer) * s->channels);
112        if (!s->decode_buffer)
113            return AVERROR(ENOMEM);
114    } else
115        s->decode_buffer = NULL;
116    s->ch_ctx = av_malloc_array(avctx->ch_layout.nb_channels, sizeof(*s->ch_ctx));
117    if (!s->ch_ctx)
118        return AVERROR(ENOMEM);
119
120    return 0;
121}
122
123static av_cold int tta_decode_init(AVCodecContext * avctx)
124{
125    TTAContext *s = avctx->priv_data;
126    GetBitContext gb;
127    int total_frames;
128    int ret;
129
130    s->avctx = avctx;
131
132    // 22 bytes for a TTA1 header
133    if (avctx->extradata_size < 22)
134        return AVERROR_INVALIDDATA;
135
136    s->crc_table = av_crc_get_table(AV_CRC_32_IEEE_LE);
137    ret = init_get_bits8(&gb, avctx->extradata, avctx->extradata_size);
138    if (ret < 0)
139        return ret;
140
141    if (show_bits_long(&gb, 32) == AV_RL32("TTA1")) {
142        /* signature */
143        skip_bits_long(&gb, 32);
144
145        s->format = get_bits(&gb, 16);
146        if (s->format > 2) {
147            av_log(avctx, AV_LOG_ERROR, "Invalid format\n");
148            return AVERROR_INVALIDDATA;
149        }
150        if (s->format == FORMAT_ENCRYPTED) {
151            if (!s->pass) {
152                av_log(avctx, AV_LOG_ERROR, "Missing password for encrypted stream. Please use the -password option\n");
153                return AVERROR(EINVAL);
154            }
155            AV_WL64(s->crc_pass, tta_check_crc64(s->pass));
156        }
157
158        s->channels = get_bits(&gb, 16);
159
160        av_channel_layout_uninit(&avctx->ch_layout);
161        if (s->channels > 1 && s->channels < 9) {
162            av_channel_layout_from_mask(&avctx->ch_layout, tta_channel_layouts[s->channels-2]);
163        } else {
164            avctx->ch_layout.order       = AV_CHANNEL_ORDER_UNSPEC;
165            avctx->ch_layout.nb_channels = s->channels;
166        }
167
168        avctx->bits_per_raw_sample = get_bits(&gb, 16);
169        s->bps = (avctx->bits_per_raw_sample + 7) / 8;
170        avctx->sample_rate = get_bits_long(&gb, 32);
171        s->data_length = get_bits_long(&gb, 32);
172        skip_bits_long(&gb, 32); // CRC32 of header
173
174        if (s->channels == 0 || s->channels > 16) {
175            av_log(avctx, AV_LOG_ERROR, "Invalid number of channels\n");
176            return AVERROR_INVALIDDATA;
177        } else if (avctx->sample_rate == 0) {
178            av_log(avctx, AV_LOG_ERROR, "Invalid samplerate\n");
179            return AVERROR_INVALIDDATA;
180        }
181
182        switch(s->bps) {
183        case 1: avctx->sample_fmt = AV_SAMPLE_FMT_U8; break;
184        case 2:
185            avctx->sample_fmt = AV_SAMPLE_FMT_S16;
186            break;
187        case 3:
188            avctx->sample_fmt = AV_SAMPLE_FMT_S32;
189            break;
190        //case 4: avctx->sample_fmt = AV_SAMPLE_FMT_S32; break;
191        default:
192            av_log(avctx, AV_LOG_ERROR, "Invalid/unsupported sample format.\n");
193            return AVERROR_INVALIDDATA;
194        }
195
196        // prevent overflow
197        if (avctx->sample_rate > 0x7FFFFFu) {
198            av_log(avctx, AV_LOG_ERROR, "sample_rate too large\n");
199            return AVERROR(EINVAL);
200        }
201        s->frame_length = 256 * avctx->sample_rate / 245;
202
203        s->last_frame_length = s->data_length % s->frame_length;
204        total_frames = s->data_length / s->frame_length +
205                       (s->last_frame_length ? 1 : 0);
206
207        av_log(avctx, AV_LOG_DEBUG, "format: %d chans: %d bps: %d rate: %d block: %d\n",
208            s->format, avctx->ch_layout.nb_channels, avctx->bits_per_coded_sample, avctx->sample_rate,
209            avctx->block_align);
210        av_log(avctx, AV_LOG_DEBUG, "data_length: %d frame_length: %d last: %d total: %d\n",
211            s->data_length, s->frame_length, s->last_frame_length, total_frames);
212
213        if(s->frame_length >= UINT_MAX / (s->channels * sizeof(int32_t))){
214            av_log(avctx, AV_LOG_ERROR, "frame_length too large\n");
215            return AVERROR_INVALIDDATA;
216        }
217    } else {
218        av_log(avctx, AV_LOG_ERROR, "Wrong extradata present\n");
219        return AVERROR_INVALIDDATA;
220    }
221
222    ff_ttadsp_init(&s->dsp);
223
224    return allocate_buffers(avctx);
225}
226
227static int tta_decode_frame(AVCodecContext *avctx, AVFrame *frame,
228                            int *got_frame_ptr, AVPacket *avpkt)
229{
230    const uint8_t *buf = avpkt->data;
231    int buf_size = avpkt->size;
232    TTAContext *s = avctx->priv_data;
233    GetBitContext gb;
234    int i, ret;
235    int cur_chan = 0, framelen = s->frame_length;
236    uint32_t *p;
237
238    if (avctx->err_recognition & AV_EF_CRCCHECK) {
239        if (buf_size < 4 ||
240            (tta_check_crc(s, buf, buf_size - 4) && avctx->err_recognition & AV_EF_EXPLODE))
241            return AVERROR_INVALIDDATA;
242    }
243
244    if ((ret = init_get_bits8(&gb, avpkt->data, avpkt->size)) < 0)
245        return ret;
246
247    /* get output buffer */
248    frame->nb_samples = framelen;
249    if ((ret = ff_thread_get_buffer(avctx, frame, 0)) < 0)
250        return ret;
251
252    // decode directly to output buffer for 24-bit sample format
253    if (s->bps == 3)
254        s->decode_buffer = (int32_t *)frame->data[0];
255
256    // init per channel states
257    for (i = 0; i < s->channels; i++) {
258        TTAFilter *filter = &s->ch_ctx[i].filter;
259        s->ch_ctx[i].predictor = 0;
260        ff_tta_filter_init(filter, ff_tta_filter_configs[s->bps-1]);
261        if (s->format == FORMAT_ENCRYPTED) {
262            int i;
263            for (i = 0; i < 8; i++)
264                filter->qm[i] = sign_extend(s->crc_pass[i], 8);
265        }
266        ff_tta_rice_init(&s->ch_ctx[i].rice, 10, 10);
267    }
268
269    i = 0;
270    for (p = s->decode_buffer; (int32_t*)p < s->decode_buffer + (framelen * s->channels); p++) {
271        int32_t *predictor = &s->ch_ctx[cur_chan].predictor;
272        TTAFilter *filter = &s->ch_ctx[cur_chan].filter;
273        TTARice *rice = &s->ch_ctx[cur_chan].rice;
274        uint32_t unary, depth, k;
275        int32_t value;
276
277        unary = get_unary(&gb, 0, get_bits_left(&gb));
278
279        if (unary == 0) {
280            depth = 0;
281            k = rice->k0;
282        } else {
283            depth = 1;
284            k = rice->k1;
285            unary--;
286        }
287
288        if (get_bits_left(&gb) < k) {
289            ret = AVERROR_INVALIDDATA;
290            goto error;
291        }
292
293        if (k) {
294            if (k > MIN_CACHE_BITS || unary > INT32_MAX >> k) {
295                ret = AVERROR_INVALIDDATA;
296                goto error;
297            }
298            value = (unary << k) + get_bits(&gb, k);
299        } else
300            value = unary;
301
302        // FIXME: copy paste from original
303        switch (depth) {
304        case 1:
305            rice->sum1 += value - (rice->sum1 >> 4);
306            if (rice->k1 > 0 && rice->sum1 < ff_tta_shift_16[rice->k1])
307                rice->k1--;
308            else if(rice->sum1 > ff_tta_shift_16[rice->k1 + 1])
309                rice->k1++;
310            value += ff_tta_shift_1[rice->k0];
311        default:
312            rice->sum0 += value - (rice->sum0 >> 4);
313            if (rice->k0 > 0 && rice->sum0 < ff_tta_shift_16[rice->k0])
314                rice->k0--;
315            else if(rice->sum0 > ff_tta_shift_16[rice->k0 + 1])
316                rice->k0++;
317        }
318
319        // extract coded value
320        *p = 1 + ((value >> 1) ^ ((value & 1) - 1));
321
322        // run hybrid filter
323        s->dsp.filter_process(filter->qm, filter->dx, filter->dl, &filter->error, p,
324                              filter->shift, filter->round);
325
326        // fixed order prediction
327#define PRED(x, k) (int32_t)((((uint64_t)(x) << (k)) - (x)) >> (k))
328        switch (s->bps) {
329        case 1: *p += PRED(*predictor, 4); break;
330        case 2:
331        case 3: *p += PRED(*predictor, 5); break;
332        case 4: *p +=      *predictor;     break;
333        }
334        *predictor = *p;
335
336        // flip channels
337        if (cur_chan < (s->channels-1))
338            cur_chan++;
339        else {
340            // decorrelate in case of multiple channels
341            if (s->channels > 1) {
342                int32_t *r = p - 1;
343                for (*p += *r / 2; r > (int32_t*)p - s->channels; r--)
344                    *r = *(r + 1) - (unsigned)*r;
345            }
346            cur_chan = 0;
347            i++;
348            // check for last frame
349            if (i == s->last_frame_length && get_bits_left(&gb) / 8 == 4) {
350                frame->nb_samples = framelen = s->last_frame_length;
351                break;
352            }
353        }
354    }
355
356    align_get_bits(&gb);
357    if (get_bits_left(&gb) < 32) {
358        ret = AVERROR_INVALIDDATA;
359        goto error;
360    }
361    skip_bits_long(&gb, 32); // frame crc
362
363    // convert to output buffer
364    switch (s->bps) {
365    case 1: {
366        uint8_t *samples = (uint8_t *)frame->data[0];
367        for (p = s->decode_buffer; (int32_t*)p < s->decode_buffer + (framelen * s->channels); p++)
368            *samples++ = *p + 0x80;
369        break;
370        }
371    case 2: {
372        int16_t *samples = (int16_t *)frame->data[0];
373        for (p = s->decode_buffer; (int32_t*)p < s->decode_buffer + (framelen * s->channels); p++)
374            *samples++ = *p;
375        break;
376        }
377    case 3: {
378        // shift samples for 24-bit sample format
379        int32_t *samples = (int32_t *)frame->data[0];
380        int overflow = 0;
381
382        for (i = 0; i < framelen * s->channels; i++) {
383            int scaled = *samples * 256U;
384            overflow += (scaled >> 8 != *samples);
385            *samples++ = scaled;
386        }
387        if (overflow)
388            av_log(avctx, AV_LOG_WARNING, "%d overflows occurred on 24bit upscale\n", overflow);
389        // reset decode buffer
390        s->decode_buffer = NULL;
391        break;
392        }
393    }
394
395    *got_frame_ptr = 1;
396
397    return buf_size;
398error:
399    // reset decode buffer
400    if (s->bps == 3)
401        s->decode_buffer = NULL;
402    return ret;
403}
404
405static av_cold int tta_decode_close(AVCodecContext *avctx) {
406    TTAContext *s = avctx->priv_data;
407
408    if (s->bps < 3)
409        av_freep(&s->decode_buffer);
410    s->decode_buffer = NULL;
411    av_freep(&s->ch_ctx);
412
413    return 0;
414}
415
416#define OFFSET(x) offsetof(TTAContext, x)
417#define DEC (AV_OPT_FLAG_DECODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM)
418static const AVOption options[] = {
419    { "password", "Set decoding password", OFFSET(pass), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, DEC },
420    { NULL },
421};
422
423static const AVClass tta_decoder_class = {
424    .class_name = "TTA Decoder",
425    .item_name  = av_default_item_name,
426    .option     = options,
427    .version    = LIBAVUTIL_VERSION_INT,
428};
429
430const FFCodec ff_tta_decoder = {
431    .p.name         = "tta",
432    .p.long_name    = NULL_IF_CONFIG_SMALL("TTA (True Audio)"),
433    .p.type         = AVMEDIA_TYPE_AUDIO,
434    .p.id           = AV_CODEC_ID_TTA,
435    .priv_data_size = sizeof(TTAContext),
436    .init           = tta_decode_init,
437    .close          = tta_decode_close,
438    FF_CODEC_DECODE_CB(tta_decode_frame),
439    .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS | AV_CODEC_CAP_CHANNEL_CONF,
440    .p.priv_class   = &tta_decoder_class,
441    .caps_internal  = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_INIT_CLEANUP,
442};
443