LCOV - code coverage report
Current view: top level - src/libavcodec - alacenc.c (source / functions) Hit Total Coverage
Test: coverage.info Lines: 292 330 88.5 %
Date: 2017-06-23 19:00:59 Functions: 14 14 100.0 %

          Line data    Source code
       1             : /*
       2             :  * ALAC audio encoder
       3             :  * Copyright (c) 2008  Jaikrishnan Menon <realityman@gmx.net>
       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             : #include "libavutil/opt.h"
      23             : 
      24             : #include "avcodec.h"
      25             : #include "put_bits.h"
      26             : #include "internal.h"
      27             : #include "lpc.h"
      28             : #include "mathops.h"
      29             : #include "alac_data.h"
      30             : 
      31             : #define DEFAULT_FRAME_SIZE        4096
      32             : #define ALAC_EXTRADATA_SIZE       36
      33             : #define ALAC_FRAME_HEADER_SIZE    55
      34             : #define ALAC_FRAME_FOOTER_SIZE    3
      35             : 
      36             : #define ALAC_ESCAPE_CODE          0x1FF
      37             : #define ALAC_MAX_LPC_ORDER        30
      38             : #define DEFAULT_MAX_PRED_ORDER    6
      39             : #define DEFAULT_MIN_PRED_ORDER    4
      40             : #define ALAC_MAX_LPC_PRECISION    9
      41             : #define ALAC_MIN_LPC_SHIFT        0
      42             : #define ALAC_MAX_LPC_SHIFT        9
      43             : 
      44             : #define ALAC_CHMODE_LEFT_RIGHT    0
      45             : #define ALAC_CHMODE_LEFT_SIDE     1
      46             : #define ALAC_CHMODE_RIGHT_SIDE    2
      47             : #define ALAC_CHMODE_MID_SIDE      3
      48             : 
      49             : typedef struct RiceContext {
      50             :     int history_mult;
      51             :     int initial_history;
      52             :     int k_modifier;
      53             :     int rice_modifier;
      54             : } RiceContext;
      55             : 
      56             : typedef struct AlacLPCContext {
      57             :     int lpc_order;
      58             :     int lpc_coeff[ALAC_MAX_LPC_ORDER+1];
      59             :     int lpc_quant;
      60             : } AlacLPCContext;
      61             : 
      62             : typedef struct AlacEncodeContext {
      63             :     const AVClass *class;
      64             :     AVCodecContext *avctx;
      65             :     int frame_size;                     /**< current frame size               */
      66             :     int verbatim;                       /**< current frame verbatim mode flag */
      67             :     int compression_level;
      68             :     int min_prediction_order;
      69             :     int max_prediction_order;
      70             :     int max_coded_frame_size;
      71             :     int write_sample_size;
      72             :     int extra_bits;
      73             :     int32_t sample_buf[2][DEFAULT_FRAME_SIZE];
      74             :     int32_t predictor_buf[2][DEFAULT_FRAME_SIZE];
      75             :     int interlacing_shift;
      76             :     int interlacing_leftweight;
      77             :     PutBitContext pbctx;
      78             :     RiceContext rc;
      79             :     AlacLPCContext lpc[2];
      80             :     LPCContext lpc_ctx;
      81             : } AlacEncodeContext;
      82             : 
      83             : 
      84        1781 : static void init_sample_buffers(AlacEncodeContext *s, int channels,
      85             :                                 const uint8_t *samples[2])
      86             : {
      87             :     int ch, i;
      88        3562 :     int shift = av_get_bytes_per_sample(s->avctx->sample_fmt) * 8 -
      89        1781 :                 s->avctx->bits_per_raw_sample;
      90             : 
      91             : #define COPY_SAMPLES(type) do {                             \
      92             :         for (ch = 0; ch < channels; ch++) {                 \
      93             :             int32_t       *bptr = s->sample_buf[ch];        \
      94             :             const type *sptr = (const type *)samples[ch];   \
      95             :             for (i = 0; i < s->frame_size; i++)             \
      96             :                 bptr[i] = sptr[i] >> shift;                 \
      97             :         }                                                   \
      98             :     } while (0)
      99             : 
     100        1781 :     if (s->avctx->sample_fmt == AV_SAMPLE_FMT_S32P)
     101        1407 :         COPY_SAMPLES(int32_t);
     102             :     else
     103         374 :         COPY_SAMPLES(int16_t);
     104        1781 : }
     105             : 
     106    14430194 : static void encode_scalar(AlacEncodeContext *s, int x,
     107             :                           int k, int write_sample_size)
     108             : {
     109             :     int divisor, q, r;
     110             : 
     111    14430194 :     k = FFMIN(k, s->rc.k_modifier);
     112    14430194 :     divisor = (1<<k) - 1;
     113    14430194 :     q = x / divisor;
     114    14430194 :     r = x % divisor;
     115             : 
     116    14430194 :     if (q > 8) {
     117             :         // write escape code and sample value directly
     118       25847 :         put_bits(&s->pbctx, 9, ALAC_ESCAPE_CODE);
     119       25847 :         put_bits(&s->pbctx, write_sample_size, x);
     120             :     } else {
     121    14404347 :         if (q)
     122     7865003 :             put_bits(&s->pbctx, q, (1<<q) - 1);
     123    14404347 :         put_bits(&s->pbctx, 1, 0);
     124             : 
     125    14404347 :         if (k != 1) {
     126    14401767 :             if (r > 0)
     127    12874329 :                 put_bits(&s->pbctx, k, r+1);
     128             :             else
     129     1527438 :                 put_bits(&s->pbctx, k-1, 0);
     130             :         }
     131             :     }
     132    14430194 : }
     133             : 
     134        2353 : static void write_element_header(AlacEncodeContext *s,
     135             :                                  enum AlacRawDataBlockType element,
     136             :                                  int instance)
     137             : {
     138        2353 :     int encode_fs = 0;
     139             : 
     140        2353 :     if (s->frame_size < DEFAULT_FRAME_SIZE)
     141           9 :         encode_fs = 1;
     142             : 
     143        2353 :     put_bits(&s->pbctx, 3,  element);               // element type
     144        2353 :     put_bits(&s->pbctx, 4,  instance);              // element instance
     145        2353 :     put_bits(&s->pbctx, 12, 0);                     // unused header bits
     146        2353 :     put_bits(&s->pbctx, 1,  encode_fs);             // Sample count is in the header
     147        2353 :     put_bits(&s->pbctx, 2,  s->extra_bits >> 3);    // Extra bytes (for 24-bit)
     148        2353 :     put_bits(&s->pbctx, 1,  s->verbatim);           // Audio block is verbatim
     149        2353 :     if (encode_fs)
     150           9 :         put_bits32(&s->pbctx, s->frame_size);       // No. of samples in the frame
     151        2353 : }
     152             : 
     153        3562 : static void calc_predictor_params(AlacEncodeContext *s, int ch)
     154             : {
     155             :     int32_t coefs[MAX_LPC_ORDER][MAX_LPC_ORDER];
     156             :     int shift[MAX_LPC_ORDER];
     157             :     int opt_order;
     158             : 
     159        3562 :     if (s->compression_level == 1) {
     160        1274 :         s->lpc[ch].lpc_order = 6;
     161        1274 :         s->lpc[ch].lpc_quant = 6;
     162        1274 :         s->lpc[ch].lpc_coeff[0] =  160;
     163        1274 :         s->lpc[ch].lpc_coeff[1] = -190;
     164        1274 :         s->lpc[ch].lpc_coeff[2] =  170;
     165        1274 :         s->lpc[ch].lpc_coeff[3] = -130;
     166        1274 :         s->lpc[ch].lpc_coeff[4] =   80;
     167        1274 :         s->lpc[ch].lpc_coeff[5] =  -25;
     168             :     } else {
     169        2288 :         opt_order = ff_lpc_calc_coefs(&s->lpc_ctx, s->sample_buf[ch],
     170             :                                       s->frame_size,
     171             :                                       s->min_prediction_order,
     172             :                                       s->max_prediction_order,
     173             :                                       ALAC_MAX_LPC_PRECISION, coefs, shift,
     174             :                                       FF_LPC_TYPE_LEVINSON, 0,
     175             :                                       ORDER_METHOD_EST, ALAC_MIN_LPC_SHIFT,
     176             :                                       ALAC_MAX_LPC_SHIFT, 1);
     177             : 
     178        2288 :         s->lpc[ch].lpc_order = opt_order;
     179        2288 :         s->lpc[ch].lpc_quant = shift[opt_order-1];
     180        2288 :         memcpy(s->lpc[ch].lpc_coeff, coefs[opt_order-1], opt_order*sizeof(int));
     181             :     }
     182        3562 : }
     183             : 
     184        1781 : static int estimate_stereo_mode(int32_t *left_ch, int32_t *right_ch, int n)
     185             : {
     186             :     int i, best;
     187             :     int32_t lt, rt;
     188             :     uint64_t sum[4];
     189             :     uint64_t score[4];
     190             : 
     191             :     /* calculate sum of 2nd order residual for each channel */
     192        1781 :     sum[0] = sum[1] = sum[2] = sum[3] = 0;
     193     7279669 :     for (i = 2; i < n; i++) {
     194     7277888 :         lt =  left_ch[i] - 2 *  left_ch[i - 1] +  left_ch[i - 2];
     195     7277888 :         rt = right_ch[i] - 2 * right_ch[i - 1] + right_ch[i - 2];
     196     7277888 :         sum[2] += FFABS((lt + rt) >> 1);
     197     7277888 :         sum[3] += FFABS(lt - rt);
     198     7277888 :         sum[0] += FFABS(lt);
     199     7277888 :         sum[1] += FFABS(rt);
     200             :     }
     201             : 
     202             :     /* calculate score for each mode */
     203        1781 :     score[0] = sum[0] + sum[1];
     204        1781 :     score[1] = sum[0] + sum[3];
     205        1781 :     score[2] = sum[1] + sum[3];
     206        1781 :     score[3] = sum[2] + sum[3];
     207             : 
     208             :     /* return mode with lowest score */
     209        1781 :     best = 0;
     210        7124 :     for (i = 1; i < 4; i++) {
     211        5343 :         if (score[i] < score[best])
     212         597 :             best = i;
     213             :     }
     214        1781 :     return best;
     215             : }
     216             : 
     217        1781 : static void alac_stereo_decorrelation(AlacEncodeContext *s)
     218             : {
     219        1781 :     int32_t *left = s->sample_buf[0], *right = s->sample_buf[1];
     220        1781 :     int i, mode, n = s->frame_size;
     221             :     int32_t tmp;
     222             : 
     223        1781 :     mode = estimate_stereo_mode(left, right, n);
     224             : 
     225        1781 :     switch (mode) {
     226        1425 :     case ALAC_CHMODE_LEFT_RIGHT:
     227        1425 :         s->interlacing_leftweight = 0;
     228        1425 :         s->interlacing_shift      = 0;
     229        1425 :         break;
     230         145 :     case ALAC_CHMODE_LEFT_SIDE:
     231      592425 :         for (i = 0; i < n; i++)
     232      592280 :             right[i] = left[i] - right[i];
     233         145 :         s->interlacing_leftweight = 1;
     234         145 :         s->interlacing_shift      = 0;
     235         145 :         break;
     236          90 :     case ALAC_CHMODE_RIGHT_SIDE:
     237      368730 :         for (i = 0; i < n; i++) {
     238      368640 :             tmp = right[i];
     239      368640 :             right[i] = left[i] - right[i];
     240      368640 :             left[i]  = tmp + (right[i] >> 31);
     241             :         }
     242          90 :         s->interlacing_leftweight = 1;
     243          90 :         s->interlacing_shift      = 31;
     244          90 :         break;
     245         121 :     default:
     246      486923 :         for (i = 0; i < n; i++) {
     247      486802 :             tmp = left[i];
     248      486802 :             left[i]  = (tmp + right[i]) >> 1;
     249      486802 :             right[i] =  tmp - right[i];
     250             :         }
     251         121 :         s->interlacing_leftweight = 1;
     252         121 :         s->interlacing_shift      = 1;
     253         121 :         break;
     254             :     }
     255        1781 : }
     256             : 
     257        3562 : static void alac_linear_predictor(AlacEncodeContext *s, int ch)
     258             : {
     259             :     int i;
     260        3562 :     AlacLPCContext lpc = s->lpc[ch];
     261        3562 :     int32_t *residual = s->predictor_buf[ch];
     262             : 
     263        3562 :     if (lpc.lpc_order == 31) {
     264           0 :         residual[0] = s->sample_buf[ch][0];
     265             : 
     266           0 :         for (i = 1; i < s->frame_size; i++) {
     267           0 :             residual[i] = s->sample_buf[ch][i    ] -
     268           0 :                           s->sample_buf[ch][i - 1];
     269             :         }
     270             : 
     271           0 :         return;
     272             :     }
     273             : 
     274             :     // generalised linear predictor
     275             : 
     276        3562 :     if (lpc.lpc_order > 0) {
     277        3562 :         int32_t *samples  = s->sample_buf[ch];
     278             : 
     279             :         // generate warm-up samples
     280        3562 :         residual[0] = samples[0];
     281       34759 :         for (i = 1; i <= lpc.lpc_order; i++)
     282       31197 :             residual[i] = sign_extend(samples[i] - samples[i-1], s->write_sample_size);
     283             : 
     284             :         // perform lpc on remaining samples
     285    14531703 :         for (i = lpc.lpc_order + 1; i < s->frame_size; i++) {
     286    14528141 :             int sum = 1 << (lpc.lpc_quant - 1), res_val, j;
     287             : 
     288   141644105 :             for (j = 0; j < lpc.lpc_order; j++) {
     289   254231928 :                 sum += (samples[lpc.lpc_order-j] - samples[0]) *
     290   127115964 :                        lpc.lpc_coeff[j];
     291             :             }
     292             : 
     293    14528141 :             sum >>= lpc.lpc_quant;
     294    14528141 :             sum += samples[0];
     295    14528141 :             residual[i] = sign_extend(samples[lpc.lpc_order+1] - sum,
     296    14528141 :                                       s->write_sample_size);
     297    14528141 :             res_val = residual[i];
     298             : 
     299    14528141 :             if (res_val) {
     300    13747876 :                 int index = lpc.lpc_order - 1;
     301    13747876 :                 int neg = (res_val < 0);
     302             : 
     303    87959314 :                 while (index >= 0 && (neg ? (res_val < 0) : (res_val > 0))) {
     304    60463562 :                     int val  = samples[0] - samples[lpc.lpc_order - index];
     305    60463562 :                     int sign = (val ? FFSIGN(val) : 0);
     306             : 
     307    60463562 :                     if (neg)
     308    28132311 :                         sign *= -1;
     309             : 
     310    60463562 :                     lpc.lpc_coeff[index] -= sign;
     311    60463562 :                     val *= sign;
     312    60463562 :                     res_val -= (val >> lpc.lpc_quant) * (lpc.lpc_order - index);
     313    60463562 :                     index--;
     314             :                 }
     315             :             }
     316    14528141 :             samples++;
     317             :         }
     318             :     }
     319             : }
     320             : 
     321        3562 : static void alac_entropy_coder(AlacEncodeContext *s, int ch)
     322             : {
     323        3562 :     unsigned int history = s->rc.initial_history;
     324        3562 :     int sign_modifier = 0, i, k;
     325        3562 :     int32_t *samples = s->predictor_buf[ch];
     326             : 
     327    14437187 :     for (i = 0; i < s->frame_size;) {
     328             :         int x;
     329             : 
     330    14430063 :         k = av_log2((history >> 9) + 3);
     331             : 
     332    14430063 :         x  = -2 * (*samples) -1;
     333    14430063 :         x ^= x >> 31;
     334             : 
     335    14430063 :         samples++;
     336    14430063 :         i++;
     337             : 
     338    14430063 :         encode_scalar(s, x - sign_modifier, k, s->write_sample_size);
     339             : 
     340    28860126 :         history += x * s->rc.history_mult -
     341    14430063 :                    ((history * s->rc.history_mult) >> 9);
     342             : 
     343    14430063 :         sign_modifier = 0;
     344    14430063 :         if (x > 0xFFFF)
     345       11363 :             history = 0xFFFF;
     346             : 
     347    14430063 :         if (history < 128 && i < s->frame_size) {
     348         131 :             unsigned int block_size = 0;
     349             : 
     350         131 :             k = 7 - av_log2(history) + ((history + 16) >> 6);
     351             : 
     352      133099 :             while (*samples == 0 && i < s->frame_size) {
     353      132837 :                 samples++;
     354      132837 :                 i++;
     355      132837 :                 block_size++;
     356             :             }
     357         131 :             encode_scalar(s, block_size, k, 16);
     358         131 :             sign_modifier = (block_size <= 0xFFFF);
     359         131 :             history = 0;
     360             :         }
     361             : 
     362             :     }
     363        3562 : }
     364             : 
     365        2353 : static void write_element(AlacEncodeContext *s,
     366             :                           enum AlacRawDataBlockType element, int instance,
     367             :                           const uint8_t *samples0, const uint8_t *samples1)
     368             : {
     369        2353 :     const uint8_t *samples[2] = { samples0, samples1 };
     370             :     int i, j, channels;
     371        2353 :     int prediction_type = 0;
     372        2353 :     PutBitContext *pb = &s->pbctx;
     373             : 
     374        2353 :     channels = element == TYPE_CPE ? 2 : 1;
     375             : 
     376        2353 :     if (s->verbatim) {
     377         572 :         write_element_header(s, element, instance);
     378             :         /* samples are channel-interleaved in verbatim mode */
     379         572 :         if (s->avctx->sample_fmt == AV_SAMPLE_FMT_S32P) {
     380         469 :             int shift = 32 - s->avctx->bits_per_raw_sample;
     381         469 :             const int32_t *samples_s32[2] = { (const int32_t *)samples0,
     382             :                                               (const int32_t *)samples1 };
     383     1920469 :             for (i = 0; i < s->frame_size; i++)
     384     5760000 :                 for (j = 0; j < channels; j++)
     385     3840000 :                     put_sbits(pb, s->avctx->bits_per_raw_sample,
     386     3840000 :                               samples_s32[j][i] >> shift);
     387             :         } else {
     388         103 :             const int16_t *samples_s16[2] = { (const int16_t *)samples0,
     389             :                                               (const int16_t *)samples1 };
     390      419053 :             for (i = 0; i < s->frame_size; i++)
     391     1256850 :                 for (j = 0; j < channels; j++)
     392      837900 :                     put_sbits(pb, s->avctx->bits_per_raw_sample,
     393      837900 :                               samples_s16[j][i]);
     394             :         }
     395             :     } else {
     396        3562 :         s->write_sample_size = s->avctx->bits_per_raw_sample - s->extra_bits +
     397        1781 :                                channels - 1;
     398             : 
     399        1781 :         init_sample_buffers(s, channels, samples);
     400        1781 :         write_element_header(s, element, instance);
     401             : 
     402             :         // extract extra bits if needed
     403        1781 :         if (s->extra_bits) {
     404        1407 :             uint32_t mask = (1 << s->extra_bits) - 1;
     405        4221 :             for (j = 0; j < channels; j++) {
     406        2814 :                 int32_t *extra = s->predictor_buf[j];
     407        2814 :                 int32_t *smp   = s->sample_buf[j];
     408    11522814 :                 for (i = 0; i < s->frame_size; i++) {
     409    11520000 :                     extra[i] = smp[i] & mask;
     410    11520000 :                     smp[i] >>= s->extra_bits;
     411             :                 }
     412             :             }
     413             :         }
     414             : 
     415        1781 :         if (channels == 2)
     416        1781 :             alac_stereo_decorrelation(s);
     417             :         else
     418           0 :             s->interlacing_shift = s->interlacing_leftweight = 0;
     419        1781 :         put_bits(pb, 8, s->interlacing_shift);
     420        1781 :         put_bits(pb, 8, s->interlacing_leftweight);
     421             : 
     422        5343 :         for (i = 0; i < channels; i++) {
     423        3562 :             calc_predictor_params(s, i);
     424             : 
     425        3562 :             put_bits(pb, 4, prediction_type);
     426        3562 :             put_bits(pb, 4, s->lpc[i].lpc_quant);
     427             : 
     428        3562 :             put_bits(pb, 3, s->rc.rice_modifier);
     429        3562 :             put_bits(pb, 5, s->lpc[i].lpc_order);
     430             :             // predictor coeff. table
     431       34759 :             for (j = 0; j < s->lpc[i].lpc_order; j++)
     432       31197 :                 put_sbits(pb, 16, s->lpc[i].lpc_coeff[j]);
     433             :         }
     434             : 
     435             :         // write extra bits if needed
     436        1781 :         if (s->extra_bits) {
     437     5761407 :             for (i = 0; i < s->frame_size; i++) {
     438    17280000 :                 for (j = 0; j < channels; j++) {
     439    11520000 :                     put_bits(pb, s->extra_bits, s->predictor_buf[j][i]);
     440             :                 }
     441             :             }
     442             :         }
     443             : 
     444             :         // apply lpc and entropy coding to audio samples
     445        5343 :         for (i = 0; i < channels; i++) {
     446        3562 :             alac_linear_predictor(s, i);
     447             : 
     448             :             // TODO: determine when this will actually help. for now it's not used.
     449        3562 :             if (prediction_type == 15) {
     450             :                 // 2nd pass 1st order filter
     451           0 :                 int32_t *residual = s->predictor_buf[i];
     452           0 :                 for (j = s->frame_size - 1; j > 0; j--)
     453           0 :                     residual[j] -= residual[j - 1];
     454             :             }
     455        3562 :             alac_entropy_coder(s, i);
     456             :         }
     457             :     }
     458        2353 : }
     459             : 
     460        2353 : static int write_frame(AlacEncodeContext *s, AVPacket *avpkt,
     461             :                        uint8_t * const *samples)
     462             : {
     463        2353 :     PutBitContext *pb = &s->pbctx;
     464        2353 :     const enum AlacRawDataBlockType *ch_elements = ff_alac_channel_elements[s->avctx->channels - 1];
     465        2353 :     const uint8_t *ch_map = ff_alac_channel_layout_offsets[s->avctx->channels - 1];
     466             :     int ch, element, sce, cpe;
     467             : 
     468        2353 :     init_put_bits(pb, avpkt->data, avpkt->size);
     469             : 
     470        2353 :     ch = element = sce = cpe = 0;
     471        7059 :     while (ch < s->avctx->channels) {
     472        2353 :         if (ch_elements[element] == TYPE_CPE) {
     473        2353 :             write_element(s, TYPE_CPE, cpe, samples[ch_map[ch]],
     474        2353 :                           samples[ch_map[ch + 1]]);
     475        2353 :             cpe++;
     476        2353 :             ch += 2;
     477             :         } else {
     478           0 :             write_element(s, TYPE_SCE, sce, samples[ch_map[ch]], NULL);
     479           0 :             sce++;
     480           0 :             ch++;
     481             :         }
     482        2353 :         element++;
     483             :     }
     484             : 
     485        2353 :     put_bits(pb, 3, TYPE_END);
     486        2353 :     flush_put_bits(pb);
     487             : 
     488        2353 :     return put_bits_count(pb) >> 3;
     489             : }
     490             : 
     491          18 : static av_always_inline int get_max_frame_size(int frame_size, int ch, int bps)
     492             : {
     493          18 :     int header_bits = 23 + 32 * (frame_size < DEFAULT_FRAME_SIZE);
     494          18 :     return FFALIGN(header_bits + bps * ch * frame_size + 3, 8) / 8;
     495             : }
     496             : 
     497           9 : static av_cold int alac_encode_close(AVCodecContext *avctx)
     498             : {
     499           9 :     AlacEncodeContext *s = avctx->priv_data;
     500           9 :     ff_lpc_end(&s->lpc_ctx);
     501           9 :     av_freep(&avctx->extradata);
     502           9 :     avctx->extradata_size = 0;
     503           9 :     return 0;
     504             : }
     505             : 
     506           9 : static av_cold int alac_encode_init(AVCodecContext *avctx)
     507             : {
     508           9 :     AlacEncodeContext *s = avctx->priv_data;
     509             :     int ret;
     510             :     uint8_t *alac_extradata;
     511             : 
     512           9 :     avctx->frame_size = s->frame_size = DEFAULT_FRAME_SIZE;
     513             : 
     514           9 :     if (avctx->sample_fmt == AV_SAMPLE_FMT_S32P) {
     515           4 :         if (avctx->bits_per_raw_sample != 24)
     516           0 :             av_log(avctx, AV_LOG_WARNING, "encoding as 24 bits-per-sample\n");
     517           4 :         avctx->bits_per_raw_sample = 24;
     518             :     } else {
     519           5 :         avctx->bits_per_raw_sample = 16;
     520           5 :         s->extra_bits              = 0;
     521             :     }
     522             : 
     523             :     // Set default compression level
     524           9 :     if (avctx->compression_level == FF_COMPRESSION_DEFAULT)
     525           2 :         s->compression_level = 2;
     526             :     else
     527           7 :         s->compression_level = av_clip(avctx->compression_level, 0, 2);
     528             : 
     529             :     // Initialize default Rice parameters
     530           9 :     s->rc.history_mult    = 40;
     531           9 :     s->rc.initial_history = 10;
     532           9 :     s->rc.k_modifier      = 14;
     533           9 :     s->rc.rice_modifier   = 4;
     534             : 
     535           9 :     s->max_coded_frame_size = get_max_frame_size(avctx->frame_size,
     536             :                                                  avctx->channels,
     537             :                                                  avctx->bits_per_raw_sample);
     538             : 
     539           9 :     avctx->extradata = av_mallocz(ALAC_EXTRADATA_SIZE + AV_INPUT_BUFFER_PADDING_SIZE);
     540           9 :     if (!avctx->extradata) {
     541           0 :         ret = AVERROR(ENOMEM);
     542           0 :         goto error;
     543             :     }
     544           9 :     avctx->extradata_size = ALAC_EXTRADATA_SIZE;
     545             : 
     546           9 :     alac_extradata = avctx->extradata;
     547           9 :     AV_WB32(alac_extradata,    ALAC_EXTRADATA_SIZE);
     548           9 :     AV_WB32(alac_extradata+4,  MKBETAG('a','l','a','c'));
     549           9 :     AV_WB32(alac_extradata+12, avctx->frame_size);
     550           9 :     AV_WB8 (alac_extradata+17, avctx->bits_per_raw_sample);
     551           9 :     AV_WB8 (alac_extradata+21, avctx->channels);
     552           9 :     AV_WB32(alac_extradata+24, s->max_coded_frame_size);
     553           9 :     AV_WB32(alac_extradata+28,
     554             :             avctx->sample_rate * avctx->channels * avctx->bits_per_raw_sample); // average bitrate
     555           9 :     AV_WB32(alac_extradata+32, avctx->sample_rate);
     556             : 
     557             :     // Set relevant extradata fields
     558           9 :     if (s->compression_level > 0) {
     559           7 :         AV_WB8(alac_extradata+18, s->rc.history_mult);
     560           7 :         AV_WB8(alac_extradata+19, s->rc.initial_history);
     561           7 :         AV_WB8(alac_extradata+20, s->rc.k_modifier);
     562             :     }
     563             : 
     564             : #if FF_API_PRIVATE_OPT
     565             : FF_DISABLE_DEPRECATION_WARNINGS
     566           9 :     if (avctx->min_prediction_order >= 0) {
     567           0 :         if (avctx->min_prediction_order < MIN_LPC_ORDER ||
     568           0 :            avctx->min_prediction_order > ALAC_MAX_LPC_ORDER) {
     569           0 :             av_log(avctx, AV_LOG_ERROR, "invalid min prediction order: %d\n",
     570             :                    avctx->min_prediction_order);
     571           0 :             ret = AVERROR(EINVAL);
     572           0 :             goto error;
     573             :         }
     574             : 
     575           0 :         s->min_prediction_order = avctx->min_prediction_order;
     576             :     }
     577             : 
     578           9 :     if (avctx->max_prediction_order >= 0) {
     579           0 :         if (avctx->max_prediction_order < MIN_LPC_ORDER ||
     580           0 :             avctx->max_prediction_order > ALAC_MAX_LPC_ORDER) {
     581           0 :             av_log(avctx, AV_LOG_ERROR, "invalid max prediction order: %d\n",
     582             :                    avctx->max_prediction_order);
     583           0 :             ret = AVERROR(EINVAL);
     584           0 :             goto error;
     585             :         }
     586             : 
     587           0 :         s->max_prediction_order = avctx->max_prediction_order;
     588             :     }
     589             : FF_ENABLE_DEPRECATION_WARNINGS
     590             : #endif
     591             : 
     592           9 :     if (s->max_prediction_order < s->min_prediction_order) {
     593           0 :         av_log(avctx, AV_LOG_ERROR,
     594             :                "invalid prediction orders: min=%d max=%d\n",
     595             :                s->min_prediction_order, s->max_prediction_order);
     596           0 :         ret = AVERROR(EINVAL);
     597           0 :         goto error;
     598             :     }
     599             : 
     600           9 :     s->avctx = avctx;
     601             : 
     602           9 :     if ((ret = ff_lpc_init(&s->lpc_ctx, avctx->frame_size,
     603             :                            s->max_prediction_order,
     604             :                            FF_LPC_TYPE_LEVINSON)) < 0) {
     605           0 :         goto error;
     606             :     }
     607             : 
     608           9 :     return 0;
     609           0 : error:
     610           0 :     alac_encode_close(avctx);
     611           0 :     return ret;
     612             : }
     613             : 
     614        2353 : static int alac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
     615             :                              const AVFrame *frame, int *got_packet_ptr)
     616             : {
     617        2353 :     AlacEncodeContext *s = avctx->priv_data;
     618             :     int out_bytes, max_frame_size, ret;
     619             : 
     620        2353 :     s->frame_size = frame->nb_samples;
     621             : 
     622        2353 :     if (frame->nb_samples < DEFAULT_FRAME_SIZE)
     623           9 :         max_frame_size = get_max_frame_size(s->frame_size, avctx->channels,
     624             :                                             avctx->bits_per_raw_sample);
     625             :     else
     626        2344 :         max_frame_size = s->max_coded_frame_size;
     627             : 
     628        2353 :     if ((ret = ff_alloc_packet2(avctx, avpkt, 4 * max_frame_size, 0)) < 0)
     629           0 :         return ret;
     630             : 
     631             :     /* use verbatim mode for compression_level 0 */
     632        2353 :     if (s->compression_level) {
     633        1781 :         s->verbatim   = 0;
     634        1781 :         s->extra_bits = avctx->bits_per_raw_sample - 16;
     635             :     } else {
     636         572 :         s->verbatim   = 1;
     637         572 :         s->extra_bits = 0;
     638             :     }
     639             : 
     640        2353 :     out_bytes = write_frame(s, avpkt, frame->extended_data);
     641             : 
     642        2353 :     if (out_bytes > max_frame_size) {
     643             :         /* frame too large. use verbatim mode */
     644           0 :         s->verbatim = 1;
     645           0 :         s->extra_bits = 0;
     646           0 :         out_bytes = write_frame(s, avpkt, frame->extended_data);
     647             :     }
     648             : 
     649        2353 :     avpkt->size = out_bytes;
     650        2353 :     *got_packet_ptr = 1;
     651        2353 :     return 0;
     652             : }
     653             : 
     654             : #define OFFSET(x) offsetof(AlacEncodeContext, x)
     655             : #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
     656             : static const AVOption options[] = {
     657             :     { "min_prediction_order", NULL, OFFSET(min_prediction_order), AV_OPT_TYPE_INT, { .i64 = DEFAULT_MIN_PRED_ORDER }, MIN_LPC_ORDER, ALAC_MAX_LPC_ORDER, AE },
     658             :     { "max_prediction_order", NULL, OFFSET(max_prediction_order), AV_OPT_TYPE_INT, { .i64 = DEFAULT_MAX_PRED_ORDER }, MIN_LPC_ORDER, ALAC_MAX_LPC_ORDER, AE },
     659             : 
     660             :     { NULL },
     661             : };
     662             : 
     663             : static const AVClass alacenc_class = {
     664             :     .class_name = "alacenc",
     665             :     .item_name  = av_default_item_name,
     666             :     .option     = options,
     667             :     .version    = LIBAVUTIL_VERSION_INT,
     668             : };
     669             : 
     670             : AVCodec ff_alac_encoder = {
     671             :     .name           = "alac",
     672             :     .long_name      = NULL_IF_CONFIG_SMALL("ALAC (Apple Lossless Audio Codec)"),
     673             :     .type           = AVMEDIA_TYPE_AUDIO,
     674             :     .id             = AV_CODEC_ID_ALAC,
     675             :     .priv_data_size = sizeof(AlacEncodeContext),
     676             :     .priv_class     = &alacenc_class,
     677             :     .init           = alac_encode_init,
     678             :     .encode2        = alac_encode_frame,
     679             :     .close          = alac_encode_close,
     680             :     .capabilities   = AV_CODEC_CAP_SMALL_LAST_FRAME,
     681             :     .channel_layouts = ff_alac_channel_layouts,
     682             :     .sample_fmts    = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S32P,
     683             :                                                      AV_SAMPLE_FMT_S16P,
     684             :                                                      AV_SAMPLE_FMT_NONE },
     685             : };

Generated by: LCOV version 1.13