libavcodec/shorten.c
85ad5695
 /*
  * Shorten decoder
  * Copyright (c) 2005 Jeff Muizelaar
  *
b78e7197
  * This file is part of FFmpeg.
  *
  * FFmpeg is free software; you can redistribute it and/or
85ad5695
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
b78e7197
  * version 2.1 of the License, or (at your option) any later version.
85ad5695
  *
b78e7197
  * FFmpeg is distributed in the hope that it will be useful,
85ad5695
  * 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
b78e7197
  * License along with FFmpeg; if not, write to the Free Software
5509bffa
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
85ad5695
  */
 
 /**
ba87f080
  * @file
85ad5695
  * Shorten decoder
  * @author Jeff Muizelaar
  *
  */
 
 #include <limits.h>
 #include "avcodec.h"
9106a698
 #include "get_bits.h"
85ad5695
 #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];
02591641
     int *coeffs;
85ad5695
     uint8_t *bitstream;
     int bitstream_size;
     int bitstream_index;
f038fe8b
     unsigned int allocated_bitstream_size;
85ad5695
     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;
 
98a6fff9
 static av_cold int shorten_decode_init(AVCodecContext * avctx)
85ad5695
 {
     ShortenContext *s = avctx->priv_data;
     s->avctx = avctx;
5d6e4c16
     avctx->sample_fmt = AV_SAMPLE_FMT_S16;
85ad5695
 
     return 0;
 }
 
3a1a7e32
 static int allocate_buffers(ShortenContext *s)
85ad5695
 {
     int i, chan;
02591641
     int *coeffs;
e1ea35fb
     void *tmp_ptr;
02591641
 
85ad5695
     for (chan=0; chan<s->channels; chan++) {
3a1a7e32
         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;
         }
 
e1ea35fb
         tmp_ptr = av_realloc(s->offset[chan], sizeof(int32_t)*FFMAX(1, s->nmean));
         if (!tmp_ptr)
             return AVERROR(ENOMEM);
         s->offset[chan] = tmp_ptr;
85ad5695
 
e1ea35fb
         tmp_ptr = av_realloc(s->decoded[chan], sizeof(int32_t)*(s->blocksize + s->nwrap));
         if (!tmp_ptr)
             return AVERROR(ENOMEM);
         s->decoded[chan] = tmp_ptr;
85ad5695
         for (i=0; i<s->nwrap; i++)
             s->decoded[chan][i] = 0;
         s->decoded[chan] += s->nwrap;
     }
02591641
 
     coeffs = av_realloc(s->coeffs, s->nwrap * sizeof(*s->coeffs));
     if (!coeffs)
         return AVERROR(ENOMEM);
     s->coeffs = coeffs;
 
3a1a7e32
     return 0;
85ad5695
 }
 
 
 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++)
107ea305
             buffer[i] <<= s->bitshift;
85ad5695
 }
 
 
 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;
 }
 
5a6a6cc7
 static inline int get_le32(GetBitContext *gb)
85ad5695
 {
8fc0162a
     return av_bswap32(get_bits_long(gb, 32));
85ad5695
 }
 
5a6a6cc7
 static inline short get_le16(GetBitContext *gb)
85ad5695
 {
8fc0162a
     return av_bswap16(get_bits_long(gb, 16));
85ad5695
 }
 
 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;
     }
 
e65ab9d9
     skip_bits_long(&hb, 32);    /* chunk_size */
85ad5695
 
     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);
dd1c8f3e
     avctx->bits_per_coded_sample = get_le16(&hb);
85ad5695
 
dd1c8f3e
     if (avctx->bits_per_coded_sample != 16) {
85ad5695
         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;
02591641
     int *coeffs = s->coeffs;
85ad5695
 
     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,
7a00bbad
         AVPacket *avpkt)
85ad5695
 {
7a00bbad
     const uint8_t *buf = avpkt->data;
     int buf_size = avpkt->size;
85ad5695
     ShortenContext *s = avctx->priv_data;
     int i, input_buf_size = 0;
     int16_t *samples = data;
     if(s->max_framesize == 0){
e1ea35fb
         void *tmp_ptr;
85ad5695
         s->max_framesize= 1024; // should hopefully be enough for the first header
e1ea35fb
         tmp_ptr = av_fast_realloc(s->bitstream, &s->allocated_bitstream_size,
                                   s->max_framesize);
         if (!tmp_ptr) {
             av_log(avctx, AV_LOG_ERROR, "error allocating bitstream buffer\n");
             return AVERROR(ENOMEM);
         }
         s->bitstream = tmp_ptr;
85ad5695
     }
 
     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){
ed186431
             *data_size = 0;
85ad5695
             return input_buf_size;
         }
     }
     init_get_bits(&s->gb, buf, buf_size*8);
7ae7300e
     skip_bits(&s->gb, s->bitindex);
85ad5695
     if (!s->blocksize)
     {
         int maxnlpc = 0;
         /* shorten signature */
a95d00b2
         if (get_bits_long(&s->gb, 32) != AV_RB32("ajkg")) {
85ad5695
             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);
 
1e25a7e7
         if (allocate_buffers(s))
             return -1;
85ad5695
 
         init_offset(s);
 
         if (s->version > 1)
             s->lpcqoffset = V2LPCQOFFSET;
 
         if (get_ur_golomb_shorten(&s->gb, FNSIZE) != FN_VERBATIM) {
64bb7bd5
             av_log(s->avctx, AV_LOG_ERROR, "missing verbatim section at beginning of stream\n");
85ad5695
             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);
02591641
                                 if (pred_order > s->nwrap) {
                                     av_log(avctx, AV_LOG_ERROR,
                                            "invalid pred_order %d\n",
                                            pred_order);
                                     return -1;
                                 }
85ad5695
                                 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) {
b08001e0
                         int out_size = s->blocksize * s->channels *
                                        av_get_bytes_per_sample(avctx->sample_fmt);
                         if (*data_size < out_size) {
                             av_log(avctx, AV_LOG_ERROR, "Output buffer is too small\n");
                             return AVERROR(EINVAL);
                         }
85ad5695
                         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;
55a96a98
             case FN_BLOCKSIZE: {
                 int blocksize = get_uint(s, av_log2(s->blocksize));
                 if (blocksize > s->blocksize) {
                     av_log(avctx, AV_LOG_ERROR, "Increasing block size is not supported\n");
                     return AVERROR_PATCHWELCOME;
                 }
                 s->blocksize = blocksize;
85ad5695
                 break;
55a96a98
             }
85ad5695
             case FN_QUIT:
ed186431
                 *data_size = 0;
85ad5695
                 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;
 }
 
98a6fff9
 static av_cold int shorten_decode_close(AVCodecContext *avctx)
85ad5695
 {
     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);
02591641
     av_freep(&s->coeffs);
85ad5695
     return 0;
 }
 
 static void shorten_flush(AVCodecContext *avctx){
     ShortenContext *s = avctx->priv_data;
 
     s->bitstream_size=
         s->bitstream_index= 0;
 }
 
e7e2df27
 AVCodec ff_shorten_decoder = {
85ad5695
     "shorten",
72415b2a
     AVMEDIA_TYPE_AUDIO,
85ad5695
     CODEC_ID_SHORTEN,
     sizeof(ShortenContext),
     shorten_decode_init,
     NULL,
     shorten_decode_close,
     shorten_decode_frame,
     .flush= shorten_flush,
fe4bf374
     .long_name= NULL_IF_CONFIG_SMALL("Shorten"),
85ad5695
 };