shorten.c 17.1 KB
Newer Older
1 2 3 4
/*
 * Shorten decoder
 * Copyright (c) 2005 Jeff Muizelaar
 *
5 6 7
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
8 9
 * modify it under the terms of the GNU Lesser General Public
 * 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 14 15 16 17
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * 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 25 26 27 28 29 30
 * Shorten decoder
 * @author Jeff Muizelaar
 *
 */

#include <limits.h>
#include "avcodec.h"
31
#include "get_bits.h"
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
#include "golomb.h"

#define MAX_CHANNELS 8
#define MAX_BLOCKSIZE 65535

#define OUT_BUFFER_SIZE 16384

#define ULONGSIZE 2

#define WAVE_FORMAT_PCM 0x0001

#define DEFAULT_BLOCK_SIZE 256

#define TYPESIZE 4
#define CHANSIZE 0
#define LPCQSIZE 2
#define ENERGYSIZE 3
#define BITSHIFTSIZE 2

#define TYPE_S16HL 3
#define TYPE_S16LH 5

#define NWRAP 3
#define NSKIPSIZE 1

#define LPCQUANT 5
#define V2LPCQOFFSET (1 << LPCQUANT)

#define FNSIZE 2
#define FN_DIFF0        0
#define FN_DIFF1        1
#define FN_DIFF2        2
#define FN_DIFF3        3
#define FN_QUIT         4
#define FN_BLOCKSIZE    5
#define FN_BITSHIFT     6
#define FN_QLPC         7
#define FN_ZERO         8
#define FN_VERBATIM     9

#define VERBATIM_CKSIZE_SIZE 5
#define VERBATIM_BYTE_SIZE 8
#define CANONICAL_HEADER_SIZE 44

typedef struct ShortenContext {
    AVCodecContext *avctx;
    GetBitContext gb;

    int min_framesize, max_framesize;
    int channels;

    int32_t *decoded[MAX_CHANNELS];
    int32_t *offset[MAX_CHANNELS];
85
    int *coeffs;
86 87 88
    uint8_t *bitstream;
    int bitstream_size;
    int bitstream_index;
89
    unsigned int allocated_bitstream_size;
90 91 92 93 94 95 96 97 98 99 100 101 102
    int header_size;
    uint8_t header[OUT_BUFFER_SIZE];
    int version;
    int cur_chan;
    int bitshift;
    int nmean;
    int internal_ftype;
    int nwrap;
    int blocksize;
    int bitindex;
    int32_t lpcqoffset;
} ShortenContext;

103
static av_cold int shorten_decode_init(AVCodecContext * avctx)
104 105 106
{
    ShortenContext *s = avctx->priv_data;
    s->avctx = avctx;
107
    avctx->sample_fmt = AV_SAMPLE_FMT_S16;
108 109 110 111

    return 0;
}

112
static int allocate_buffers(ShortenContext *s)
113 114
{
    int i, chan;
115 116
    int *coeffs;

117
    for (chan=0; chan<s->channels; chan++) {
118 119 120 121 122 123 124 125 126
        if(FFMAX(1, s->nmean) >= UINT_MAX/sizeof(int32_t)){
            av_log(s->avctx, AV_LOG_ERROR, "nmean too large\n");
            return -1;
        }
        if(s->blocksize + s->nwrap >= UINT_MAX/sizeof(int32_t) || s->blocksize + s->nwrap <= (unsigned)s->nwrap){
            av_log(s->avctx, AV_LOG_ERROR, "s->blocksize + s->nwrap too large\n");
            return -1;
        }

127 128 129 130 131 132 133
        s->offset[chan] = av_realloc(s->offset[chan], sizeof(int32_t)*FFMAX(1, s->nmean));

        s->decoded[chan] = av_realloc(s->decoded[chan], sizeof(int32_t)*(s->blocksize + s->nwrap));
        for (i=0; i<s->nwrap; i++)
            s->decoded[chan][i] = 0;
        s->decoded[chan] += s->nwrap;
    }
134 135 136 137 138 139

    coeffs = av_realloc(s->coeffs, s->nwrap * sizeof(*s->coeffs));
    if (!coeffs)
        return AVERROR(ENOMEM);
    s->coeffs = coeffs;

140
    return 0;
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183
}


static inline unsigned int get_uint(ShortenContext *s, int k)
{
    if (s->version != 0)
        k = get_ur_golomb_shorten(&s->gb, ULONGSIZE);
    return get_ur_golomb_shorten(&s->gb, k);
}


static void fix_bitshift(ShortenContext *s, int32_t *buffer)
{
    int i;

    if (s->bitshift != 0)
        for (i = 0; i < s->blocksize; i++)
            buffer[s->nwrap + i] <<= s->bitshift;
}


static void init_offset(ShortenContext *s)
{
    int32_t mean = 0;
    int  chan, i;
    int nblock = FFMAX(1, s->nmean);
    /* initialise offset */
    switch (s->internal_ftype)
    {
        case TYPE_S16HL:
        case TYPE_S16LH:
            mean = 0;
            break;
        default:
            av_log(s->avctx, AV_LOG_ERROR, "unknown audio type");
            abort();
    }

    for (chan = 0; chan < s->channels; chan++)
        for (i = 0; i < nblock; i++)
            s->offset[chan][i] = mean;
}

184
static inline int get_le32(GetBitContext *gb)
185
{
186
    return av_bswap32(get_bits_long(gb, 32));
187 188
}

189
static inline short get_le16(GetBitContext *gb)
190
{
191
    return av_bswap16(get_bits_long(gb, 16));
192 193 194 195 196 197 198 199 200 201 202 203 204 205
}

static int decode_wave_header(AVCodecContext *avctx, uint8_t *header, int header_size)
{
    GetBitContext hb;
    int len;
    short wave_format;

    init_get_bits(&hb, header, header_size*8);
    if (get_le32(&hb) != MKTAG('R','I','F','F')) {
        av_log(avctx, AV_LOG_ERROR, "missing RIFF tag\n");
        return -1;
    }

Mans Rullgard's avatar
Mans Rullgard committed
206
    skip_bits_long(&hb, 32);    /* chunk_size */
207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237

    if (get_le32(&hb) != MKTAG('W','A','V','E')) {
        av_log(avctx, AV_LOG_ERROR, "missing WAVE tag\n");
        return -1;
    }

    while (get_le32(&hb) != MKTAG('f','m','t',' ')) {
        len = get_le32(&hb);
        skip_bits(&hb, 8*len);
    }
    len = get_le32(&hb);

    if (len < 16) {
        av_log(avctx, AV_LOG_ERROR, "fmt chunk was too short\n");
        return -1;
    }

    wave_format = get_le16(&hb);

    switch (wave_format) {
        case WAVE_FORMAT_PCM:
            break;
        default:
            av_log(avctx, AV_LOG_ERROR, "unsupported wave format\n");
            return -1;
    }

    avctx->channels = get_le16(&hb);
    avctx->sample_rate = get_le32(&hb);
    avctx->bit_rate = get_le32(&hb) * 8;
    avctx->block_align = get_le16(&hb);
238
    avctx->bits_per_coded_sample = get_le16(&hb);
239

240
    if (avctx->bits_per_coded_sample != 16) {
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
        av_log(avctx, AV_LOG_ERROR, "unsupported number of bits per sample\n");
        return -1;
    }

    len -= 16;
    if (len > 0)
        av_log(avctx, AV_LOG_INFO, "%d header bytes unparsed\n", len);

    return 0;
}

static int16_t * interleave_buffer(int16_t *samples, int nchan, int blocksize, int32_t **buffer) {
    int i, chan;
    for (i=0; i<blocksize; i++)
        for (chan=0; chan < nchan; chan++)
            *samples++ = FFMIN(buffer[chan][i], 32768);
    return samples;
}

static void decode_subframe_lpc(ShortenContext *s, int channel, int residual_size, int pred_order)
{
    int sum, i, j;
263
    int *coeffs = s->coeffs;
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278

    for (i=0; i<pred_order; i++)
        coeffs[i] = get_sr_golomb_shorten(&s->gb, LPCQUANT);

    for (i=0; i < s->blocksize; i++) {
        sum = s->lpcqoffset;
        for (j=0; j<pred_order; j++)
            sum += coeffs[j] * s->decoded[channel][i-j-1];
        s->decoded[channel][i] = get_sr_golomb_shorten(&s->gb, residual_size) + (sum >> LPCQUANT);
    }
}


static int shorten_decode_frame(AVCodecContext *avctx,
        void *data, int *data_size,
279
        AVPacket *avpkt)
280
{
281 282
    const uint8_t *buf = avpkt->data;
    int buf_size = avpkt->size;
283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
    ShortenContext *s = avctx->priv_data;
    int i, input_buf_size = 0;
    int16_t *samples = data;
    if(s->max_framesize == 0){
        s->max_framesize= 1024; // should hopefully be enough for the first header
        s->bitstream= av_fast_realloc(s->bitstream, &s->allocated_bitstream_size, s->max_framesize);
    }

    if(1 && s->max_framesize){//FIXME truncated
        buf_size= FFMIN(buf_size, s->max_framesize - s->bitstream_size);
        input_buf_size= buf_size;

        if(s->bitstream_index + s->bitstream_size + buf_size > s->allocated_bitstream_size){
            //                printf("memmove\n");
            memmove(s->bitstream, &s->bitstream[s->bitstream_index], s->bitstream_size);
            s->bitstream_index=0;
        }
        memcpy(&s->bitstream[s->bitstream_index + s->bitstream_size], buf, buf_size);
        buf= &s->bitstream[s->bitstream_index];
        buf_size += s->bitstream_size;
        s->bitstream_size= buf_size;

        if(buf_size < s->max_framesize){
306
            *data_size = 0;
307 308 309 310
            return input_buf_size;
        }
    }
    init_get_bits(&s->gb, buf, buf_size*8);
311
    skip_bits(&s->gb, s->bitindex);
312 313 314 315
    if (!s->blocksize)
    {
        int maxnlpc = 0;
        /* shorten signature */
316
        if (get_bits_long(&s->gb, 32) != AV_RB32("ajkg")) {
317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347
            av_log(s->avctx, AV_LOG_ERROR, "missing shorten magic 'ajkg'\n");
            return -1;
        }

        s->lpcqoffset = 0;
        s->blocksize = DEFAULT_BLOCK_SIZE;
        s->channels = 1;
        s->nmean = -1;
        s->version = get_bits(&s->gb, 8);
        s->internal_ftype = get_uint(s, TYPESIZE);

        s->channels = get_uint(s, CHANSIZE);
        if (s->channels > MAX_CHANNELS) {
            av_log(s->avctx, AV_LOG_ERROR, "too many channels: %d\n", s->channels);
            return -1;
        }

        /* get blocksize if version > 0 */
        if (s->version > 0) {
            int skip_bytes;
            s->blocksize = get_uint(s, av_log2(DEFAULT_BLOCK_SIZE));
            maxnlpc = get_uint(s, LPCQSIZE);
            s->nmean = get_uint(s, 0);

            skip_bytes = get_uint(s, NSKIPSIZE);
            for (i=0; i<skip_bytes; i++) {
                skip_bits(&s->gb, 8);
            }
        }
        s->nwrap = FFMAX(NWRAP, maxnlpc);

348 349
        if (allocate_buffers(s))
            return -1;
350 351 352 353 354 355 356

        init_offset(s);

        if (s->version > 1)
            s->lpcqoffset = V2LPCQOFFSET;

        if (get_ur_golomb_shorten(&s->gb, FNSIZE) != FN_VERBATIM) {
357
            av_log(s->avctx, AV_LOG_ERROR, "missing verbatim section at beginning of stream\n");
358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
            return -1;
        }

        s->header_size = get_ur_golomb_shorten(&s->gb, VERBATIM_CKSIZE_SIZE);
        if (s->header_size >= OUT_BUFFER_SIZE || s->header_size < CANONICAL_HEADER_SIZE) {
            av_log(s->avctx, AV_LOG_ERROR, "header is wrong size: %d\n", s->header_size);
            return -1;
        }

        for (i=0; i<s->header_size; i++)
            s->header[i] = (char)get_ur_golomb_shorten(&s->gb, VERBATIM_BYTE_SIZE);

        if (decode_wave_header(avctx, s->header, s->header_size) < 0)
            return -1;

        s->cur_chan = 0;
        s->bitshift = 0;
    }
    else
    {
        int cmd;
        int len;
        cmd = get_ur_golomb_shorten(&s->gb, FNSIZE);
        switch (cmd) {
            case FN_ZERO:
            case FN_DIFF0:
            case FN_DIFF1:
            case FN_DIFF2:
            case FN_DIFF3:
            case FN_QLPC:
                {
                    int residual_size = 0;
                    int channel = s->cur_chan;
                    int32_t coffset;
                    if (cmd != FN_ZERO) {
                        residual_size = get_ur_golomb_shorten(&s->gb, ENERGYSIZE);
                        /* this is a hack as version 0 differed in defintion of get_sr_golomb_shorten */
                        if (s->version == 0)
                            residual_size--;
                    }

                    if (s->nmean == 0)
                        coffset = s->offset[channel][0];
                    else {
                        int32_t sum = (s->version < 2) ? 0 : s->nmean / 2;
                        for (i=0; i<s->nmean; i++)
                            sum += s->offset[channel][i];
                        coffset = sum / s->nmean;
                        if (s->version >= 2)
                            coffset >>= FFMIN(1, s->bitshift);
                    }
                    switch (cmd) {
                        case FN_ZERO:
                            for (i=0; i<s->blocksize; i++)
                                s->decoded[channel][i] = 0;
                            break;
                        case FN_DIFF0:
                            for (i=0; i<s->blocksize; i++)
                                s->decoded[channel][i] = get_sr_golomb_shorten(&s->gb, residual_size) + coffset;
                            break;
                        case FN_DIFF1:
                            for (i=0; i<s->blocksize; i++)
                                s->decoded[channel][i] = get_sr_golomb_shorten(&s->gb, residual_size) + s->decoded[channel][i - 1];
                            break;
                        case FN_DIFF2:
                            for (i=0; i<s->blocksize; i++)
                                s->decoded[channel][i] = get_sr_golomb_shorten(&s->gb, residual_size) + 2*s->decoded[channel][i-1]
                                                                                                      -   s->decoded[channel][i-2];
                            break;
                        case FN_DIFF3:
                            for (i=0; i<s->blocksize; i++)
                                s->decoded[channel][i] = get_sr_golomb_shorten(&s->gb, residual_size) + 3*s->decoded[channel][i-1]
                                                                                                      - 3*s->decoded[channel][i-2]
                                                                                                      +   s->decoded[channel][i-3];
                            break;
                        case FN_QLPC:
                            {
                                int pred_order = get_ur_golomb_shorten(&s->gb, LPCQSIZE);
436 437 438 439 440 441
                                if (pred_order > s->nwrap) {
                                    av_log(avctx, AV_LOG_ERROR,
                                           "invalid pred_order %d\n",
                                           pred_order);
                                    return -1;
                                }
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489
                                for (i=0; i<pred_order; i++)
                                    s->decoded[channel][i - pred_order] -= coffset;
                                decode_subframe_lpc(s, channel, residual_size, pred_order);
                                if (coffset != 0)
                                    for (i=0; i < s->blocksize; i++)
                                        s->decoded[channel][i] += coffset;
                            }
                    }
                    if (s->nmean > 0) {
                        int32_t sum = (s->version < 2) ? 0 : s->blocksize / 2;
                        for (i=0; i<s->blocksize; i++)
                            sum += s->decoded[channel][i];

                        for (i=1; i<s->nmean; i++)
                            s->offset[channel][i-1] = s->offset[channel][i];

                        if (s->version < 2)
                            s->offset[channel][s->nmean - 1] = sum / s->blocksize;
                        else
                            s->offset[channel][s->nmean - 1] = (sum / s->blocksize) << s->bitshift;
                    }
                    for (i=-s->nwrap; i<0; i++)
                        s->decoded[channel][i] = s->decoded[channel][i + s->blocksize];

                    fix_bitshift(s, s->decoded[channel]);

                    s->cur_chan++;
                    if (s->cur_chan == s->channels) {
                        samples = interleave_buffer(samples, s->channels, s->blocksize, s->decoded);
                        s->cur_chan = 0;
                        goto frame_done;
                    }
                    break;
                }
                break;
            case FN_VERBATIM:
                len = get_ur_golomb_shorten(&s->gb, VERBATIM_CKSIZE_SIZE);
                while (len--) {
                    get_ur_golomb_shorten(&s->gb, VERBATIM_BYTE_SIZE);
                }
                break;
            case FN_BITSHIFT:
                s->bitshift = get_ur_golomb_shorten(&s->gb, BITSHIFTSIZE);
                break;
            case FN_BLOCKSIZE:
                s->blocksize = get_uint(s, av_log2(s->blocksize));
                break;
            case FN_QUIT:
490
                *data_size = 0;
491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518
                return buf_size;
                break;
            default:
                av_log(avctx, AV_LOG_ERROR, "unknown shorten function %d\n", cmd);
                return -1;
                break;
        }
    }
frame_done:
    *data_size = (int8_t *)samples - (int8_t *)data;

    //    s->last_blocksize = s->blocksize;
    s->bitindex = get_bits_count(&s->gb) - 8*((get_bits_count(&s->gb))/8);
    i= (get_bits_count(&s->gb))/8;
    if (i > buf_size) {
        av_log(s->avctx, AV_LOG_ERROR, "overread: %d\n", i - buf_size);
        s->bitstream_size=0;
        s->bitstream_index=0;
        return -1;
    }
    if (s->bitstream_size) {
        s->bitstream_index += i;
        s->bitstream_size  -= i;
        return input_buf_size;
    } else
        return i;
}

519
static av_cold int shorten_decode_close(AVCodecContext *avctx)
520 521 522 523 524 525 526 527 528 529
{
    ShortenContext *s = avctx->priv_data;
    int i;

    for (i = 0; i < s->channels; i++) {
        s->decoded[i] -= s->nwrap;
        av_freep(&s->decoded[i]);
        av_freep(&s->offset[i]);
    }
    av_freep(&s->bitstream);
530
    av_freep(&s->coeffs);
531 532 533 534 535 536 537 538 539 540
    return 0;
}

static void shorten_flush(AVCodecContext *avctx){
    ShortenContext *s = avctx->priv_data;

    s->bitstream_size=
        s->bitstream_index= 0;
}

541
AVCodec ff_shorten_decoder = {
542
    "shorten",
543
    AVMEDIA_TYPE_AUDIO,
544 545 546 547 548 549 550
    CODEC_ID_SHORTEN,
    sizeof(ShortenContext),
    shorten_decode_init,
    NULL,
    shorten_decode_close,
    shorten_decode_frame,
    .flush= shorten_flush,
551
    .long_name= NULL_IF_CONFIG_SMALL("Shorten"),
552
};