* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
-#define ALT_BITSTREAM_READER_LE
+#include "libavutil/avassert.h"
+#include "libavutil/channel_layout.h"
+#include "libavutil/opt.h"
#include "avcodec.h"
#include "dsputil.h"
-#include "get_bits.h"
#include "bytestream.h"
-#include "libavutil/audioconvert.h"
+#include "internal.h"
/**
* @file
* Monkey's Audio lossless audio decoder
*/
-#define BLOCKS_PER_LOOP 4608
#define MAX_CHANNELS 2
#define MAX_BYTESPERSAMPLE 3
/** Decoder context */
typedef struct APEContext {
+ AVClass *class; ///< class for AVOptions
AVCodecContext *avctx;
DSPContext dsp;
int channels;
int samples; ///< samples left to decode in current frame
+ int bps;
int fileversion; ///< codec version, very important in decoding process
int compression_level; ///< compression levels
uint32_t CRC; ///< frame CRC
int frameflags; ///< frame flags
- int currentframeblocks; ///< samples (per channel) in current frame
- int blocksdecoded; ///< count of decoded samples in current frame
APEPredictor predictor; ///< predictor used for final reconstruction
- int32_t decoded0[BLOCKS_PER_LOOP]; ///< decoded data for the first channel
- int32_t decoded1[BLOCKS_PER_LOOP]; ///< decoded data for the second channel
+ int32_t *decoded_buffer;
+ int decoded_size;
+ int32_t *decoded[MAX_CHANNELS]; ///< decoded data for each channel
+ int blocks_per_loop; ///< maximum number of samples to decode for each call
int16_t* filterbuf[APE_FILTER_LEVELS]; ///< filter memory
uint8_t *data; ///< current frame data
uint8_t *data_end; ///< frame data end
+ int data_size; ///< frame data allocated size
const uint8_t *ptr; ///< current position in frame data
- const uint8_t *last_ptr; ///< position where last 4608-sample block ended
int error;
} APEContext;
for (i = 0; i < APE_FILTER_LEVELS; i++)
av_freep(&s->filterbuf[i]);
+ av_freep(&s->decoded_buffer);
av_freep(&s->data);
+ s->decoded_size = s->data_size = 0;
+
return 0;
}
av_log(avctx, AV_LOG_ERROR, "Incorrect extradata\n");
return AVERROR(EINVAL);
}
- if (avctx->bits_per_coded_sample != 16) {
- av_log(avctx, AV_LOG_ERROR, "Only 16-bit samples are supported\n");
- return AVERROR(EINVAL);
- }
if (avctx->channels > 2) {
av_log(avctx, AV_LOG_ERROR, "Only mono and stereo is supported\n");
return AVERROR(EINVAL);
}
+ s->bps = avctx->bits_per_coded_sample;
+ switch (s->bps) {
+ case 8:
+ avctx->sample_fmt = AV_SAMPLE_FMT_U8P;
+ break;
+ case 16:
+ avctx->sample_fmt = AV_SAMPLE_FMT_S16P;
+ break;
+ case 24:
+ avctx->sample_fmt = AV_SAMPLE_FMT_S32P;
+ break;
+ default:
+ av_log_ask_for_sample(avctx, "Unsupported bits per coded sample %d\n",
+ s->bps);
+ return AVERROR_PATCHWELCOME;
+ }
s->avctx = avctx;
s->channels = avctx->channels;
s->fileversion = AV_RL16(avctx->extradata);
filter_alloc_fail);
}
- dsputil_init(&s->dsp, avctx);
- avctx->sample_fmt = AV_SAMPLE_FMT_S16;
+ ff_dsputil_init(&s->dsp, avctx);
avctx->channel_layout = (avctx->channels==2) ? AV_CH_LAYOUT_STEREO : AV_CH_LAYOUT_MONO;
+
return 0;
filter_alloc_fail:
ape_decode_close(avctx);
{
while (ctx->rc.range <= BOTTOM_VALUE) {
ctx->rc.buffer <<= 8;
- if(ctx->ptr < ctx->data_end)
+ if(ctx->ptr < ctx->data_end) {
ctx->rc.buffer += *ctx->ptr;
- ctx->ptr++;
+ ctx->ptr++;
+ } else {
+ ctx->error = 1;
+ }
ctx->rc.low = (ctx->rc.low << 8) | ((ctx->rc.buffer >> 1) & 0xFF);
ctx->rc.range <<= 8;
}
}
/** @} */ // group rangecoder
-static inline void update_rice(APERice *rice, int x)
+static inline void update_rice(APERice *rice, unsigned int x)
{
int lim = rice->k ? (1 << (rice->k + 4)) : 0;
rice->ksum += ((x + 1) / 2) - ((rice->ksum + 16) >> 5);
static inline int ape_decode_value(APEContext *ctx, APERice *rice)
{
- int x, overflow;
+ unsigned int x, overflow;
if (ctx->fileversion < 3990) {
int tmpk;
if (tmpk <= 16)
x = range_decode_bits(ctx, tmpk);
- else {
+ else if (tmpk <= 32) {
x = range_decode_bits(ctx, 16);
x |= (range_decode_bits(ctx, tmpk - 16) << 16);
+ } else {
+ av_log(ctx->avctx, AV_LOG_ERROR, "Too many bits: %d\n", tmpk);
+ return AVERROR_INVALIDDATA;
}
x += overflow << tmpk;
} else {
static void entropy_decode(APEContext *ctx, int blockstodecode, int stereo)
{
- int32_t *decoded0 = ctx->decoded0;
- int32_t *decoded1 = ctx->decoded1;
+ int32_t *decoded0 = ctx->decoded[0];
+ int32_t *decoded1 = ctx->decoded[1];
- ctx->blocksdecoded = blockstodecode;
-
- if (ctx->frameflags & APE_FRAMECODE_STEREO_SILENCE) {
- /* We are pure silence, just memset the output buffer. */
- memset(decoded0, 0, blockstodecode * sizeof(int32_t));
- memset(decoded1, 0, blockstodecode * sizeof(int32_t));
- } else {
- while (blockstodecode--) {
- *decoded0++ = ape_decode_value(ctx, &ctx->riceY);
- if (stereo)
- *decoded1++ = ape_decode_value(ctx, &ctx->riceX);
- }
+ while (blockstodecode--) {
+ *decoded0++ = ape_decode_value(ctx, &ctx->riceY);
+ if (stereo)
+ *decoded1++ = ape_decode_value(ctx, &ctx->riceX);
}
-
- if (ctx->blocksdecoded == ctx->currentframeblocks)
- range_dec_normalize(ctx); /* normalize to use up all bytes */
}
-static void init_entropy_decoder(APEContext *ctx)
+static int init_entropy_decoder(APEContext *ctx)
{
/* Read the CRC */
+ if (ctx->data_end - ctx->ptr < 6)
+ return AVERROR_INVALIDDATA;
ctx->CRC = bytestream_get_be32(&ctx->ptr);
/* Read the frame flags if they exist */
if ((ctx->fileversion > 3820) && (ctx->CRC & 0x80000000)) {
ctx->CRC &= ~0x80000000;
+ if (ctx->data_end - ctx->ptr < 6)
+ return AVERROR_INVALIDDATA;
ctx->frameflags = bytestream_get_be32(&ctx->ptr);
}
- /* Keep a count of the blocks decoded in this frame */
- ctx->blocksdecoded = 0;
-
/* Initialize the rice structs */
ctx->riceX.k = 10;
ctx->riceX.ksum = (1 << ctx->riceX.k) * 16;
ctx->ptr++;
range_start_decoding(ctx);
+
+ return 0;
}
static const int32_t initial_coeffs[4] = {
APEPredictor *p = &ctx->predictor;
/* Zero the history buffers */
- memset(p->historybuffer, 0, PREDICTOR_SIZE * sizeof(int32_t));
+ memset(p->historybuffer, 0, PREDICTOR_SIZE * sizeof(*p->historybuffer));
p->buf = p->historybuffer;
/* Initialize and zero the coefficients */
static void predictor_decode_stereo(APEContext *ctx, int count)
{
APEPredictor *p = &ctx->predictor;
- int32_t *decoded0 = ctx->decoded0;
- int32_t *decoded1 = ctx->decoded1;
+ int32_t *decoded0 = ctx->decoded[0];
+ int32_t *decoded1 = ctx->decoded[1];
while (count--) {
/* Predictor Y */
/* Have we filled the history buffer? */
if (p->buf == p->historybuffer + HISTORY_SIZE) {
- memmove(p->historybuffer, p->buf, PREDICTOR_SIZE * sizeof(int32_t));
+ memmove(p->historybuffer, p->buf,
+ PREDICTOR_SIZE * sizeof(*p->historybuffer));
p->buf = p->historybuffer;
}
}
static void predictor_decode_mono(APEContext *ctx, int count)
{
APEPredictor *p = &ctx->predictor;
- int32_t *decoded0 = ctx->decoded0;
+ int32_t *decoded0 = ctx->decoded[0];
int32_t predictionA, currentA, A, sign;
currentA = p->lastA[0];
/* Have we filled the history buffer? */
if (p->buf == p->historybuffer + HISTORY_SIZE) {
- memmove(p->historybuffer, p->buf, PREDICTOR_SIZE * sizeof(int32_t));
+ memmove(p->historybuffer, p->buf,
+ PREDICTOR_SIZE * sizeof(*p->historybuffer));
p->buf = p->historybuffer;
}
f->delay = f->historybuffer + order * 2;
f->adaptcoeffs = f->historybuffer + order;
- memset(f->historybuffer, 0, (order * 2) * sizeof(int16_t));
- memset(f->coeffs, 0, order * sizeof(int16_t));
+ memset(f->historybuffer, 0, (order * 2) * sizeof(*f->historybuffer));
+ memset(f->coeffs, 0, order * sizeof(*f->coeffs));
f->avg = 0;
}
/* Update the adaption coefficients */
absres = FFABS(res);
if (absres)
- *f->adaptcoeffs = ((res & (1<<31)) - (1<<30)) >>
+ *f->adaptcoeffs = ((res & (-1<<31)) ^ (-1<<30)) >>
(25 + (absres <= f->avg*3) + (absres <= f->avg*4/3));
else
*f->adaptcoeffs = 0;
/* Have we filled the history buffer? */
if (f->delay == f->historybuffer + HISTORY_SIZE + (order * 2)) {
memmove(f->historybuffer, f->delay - (order * 2),
- (order * 2) * sizeof(int16_t));
+ (order * 2) * sizeof(*f->historybuffer));
f->delay = f->historybuffer + order * 2;
f->adaptcoeffs = f->historybuffer + order;
}
}
}
-static void init_frame_decoder(APEContext *ctx)
+static int init_frame_decoder(APEContext *ctx)
{
- int i;
- init_entropy_decoder(ctx);
+ int i, ret;
+ if ((ret = init_entropy_decoder(ctx)) < 0)
+ return ret;
init_predictor_decoder(ctx);
for (i = 0; i < APE_FILTER_LEVELS; i++) {
init_filter(ctx, ctx->filters[i], ctx->filterbuf[i],
ape_filter_orders[ctx->fset][i]);
}
+ return 0;
}
static void ape_unpack_mono(APEContext *ctx, int count)
{
- int32_t *decoded0 = ctx->decoded0;
- int32_t *decoded1 = ctx->decoded1;
-
if (ctx->frameflags & APE_FRAMECODE_STEREO_SILENCE) {
- entropy_decode(ctx, count, 0);
/* We are pure silence, so we're done. */
av_log(ctx->avctx, AV_LOG_DEBUG, "pure silence mono\n");
return;
}
entropy_decode(ctx, count, 0);
- ape_apply_filters(ctx, decoded0, NULL, count);
+ ape_apply_filters(ctx, ctx->decoded[0], NULL, count);
/* Now apply the predictor decoding */
predictor_decode_mono(ctx, count);
/* Pseudo-stereo - just copy left channel to right channel */
if (ctx->channels == 2) {
- memcpy(decoded1, decoded0, count * sizeof(*decoded1));
+ memcpy(ctx->decoded[1], ctx->decoded[0], count * sizeof(*ctx->decoded[1]));
}
}
static void ape_unpack_stereo(APEContext *ctx, int count)
{
int32_t left, right;
- int32_t *decoded0 = ctx->decoded0;
- int32_t *decoded1 = ctx->decoded1;
+ int32_t *decoded0 = ctx->decoded[0];
+ int32_t *decoded1 = ctx->decoded[1];
if (ctx->frameflags & APE_FRAMECODE_STEREO_SILENCE) {
/* We are pure silence, so we're done. */
}
}
-static int ape_decode_frame(AVCodecContext *avctx,
- void *data, int *data_size,
- AVPacket *avpkt)
+static int ape_decode_frame(AVCodecContext *avctx, void *data,
+ int *got_frame_ptr, AVPacket *avpkt)
{
+ AVFrame *frame = data;
const uint8_t *buf = avpkt->data;
- int buf_size = avpkt->size;
APEContext *s = avctx->priv_data;
- int16_t *samples = data;
- uint32_t nblocks;
- int i, n;
+ uint8_t *sample8;
+ int16_t *sample16;
+ int32_t *sample24;
+ int i, ch, ret;
int blockstodecode;
- int bytes_used;
+ int bytes_used = 0;
- /* should not happen but who knows */
- if (BLOCKS_PER_LOOP * 2 * avctx->channels > *data_size) {
- av_log (avctx, AV_LOG_ERROR, "Output buffer is too small.\n");
- return AVERROR(EINVAL);
- }
+ /* this should never be negative, but bad things will happen if it is, so
+ check it just to make sure. */
+ av_assert0(s->samples >= 0);
if(!s->samples){
- void *tmp_data = av_realloc(s->data, (buf_size + 3) & ~3);
- if (!tmp_data)
+ uint32_t nblocks, offset;
+ int buf_size;
+
+ if (!avpkt->size) {
+ *got_frame_ptr = 0;
+ return 0;
+ }
+ if (avpkt->size < 8) {
+ av_log(avctx, AV_LOG_ERROR, "Packet is too small\n");
+ return AVERROR_INVALIDDATA;
+ }
+ buf_size = avpkt->size & ~3;
+ if (buf_size != avpkt->size) {
+ av_log(avctx, AV_LOG_WARNING, "packet size is not a multiple of 4. "
+ "extra bytes at the end will be skipped.\n");
+ }
+
+ av_fast_malloc(&s->data, &s->data_size, buf_size);
+ if (!s->data)
return AVERROR(ENOMEM);
- s->data = tmp_data;
s->dsp.bswap_buf((uint32_t*)s->data, (const uint32_t*)buf, buf_size >> 2);
- s->ptr = s->last_ptr = s->data;
+ s->ptr = s->data;
s->data_end = s->data + buf_size;
nblocks = bytestream_get_be32(&s->ptr);
- n = bytestream_get_be32(&s->ptr);
- if(n < 0 || n > 3){
+ offset = bytestream_get_be32(&s->ptr);
+ if (offset > 3) {
av_log(avctx, AV_LOG_ERROR, "Incorrect offset passed\n");
s->data = NULL;
return AVERROR_INVALIDDATA;
}
- s->ptr += n;
+ if (s->data_end - s->ptr < offset) {
+ av_log(avctx, AV_LOG_ERROR, "Packet is too small\n");
+ return AVERROR_INVALIDDATA;
+ }
+ s->ptr += offset;
- buf += 4;
if (!nblocks || nblocks > INT_MAX) {
av_log(avctx, AV_LOG_ERROR, "Invalid sample count: %u.\n", nblocks);
return AVERROR_INVALIDDATA;
}
- s->currentframeblocks = s->samples = nblocks;
-
- memset(s->decoded0, 0, sizeof(s->decoded0));
- memset(s->decoded1, 0, sizeof(s->decoded1));
+ s->samples = nblocks;
/* Initialize the frame decoder */
- init_frame_decoder(s);
+ if (init_frame_decoder(s) < 0) {
+ av_log(avctx, AV_LOG_ERROR, "Error reading frame header\n");
+ return AVERROR_INVALIDDATA;
+ }
+
+ bytes_used = avpkt->size;
}
if (!s->data) {
- *data_size = 0;
- return buf_size;
+ *got_frame_ptr = 0;
+ return avpkt->size;
}
- nblocks = s->samples;
- blockstodecode = FFMIN(BLOCKS_PER_LOOP, nblocks);
+ blockstodecode = FFMIN(s->blocks_per_loop, s->samples);
+
+ /* reallocate decoded sample buffer if needed */
+ av_fast_malloc(&s->decoded_buffer, &s->decoded_size,
+ 2 * FFALIGN(blockstodecode, 8) * sizeof(*s->decoded_buffer));
+ if (!s->decoded_buffer)
+ return AVERROR(ENOMEM);
+ memset(s->decoded_buffer, 0, s->decoded_size);
+ s->decoded[0] = s->decoded_buffer;
+ s->decoded[1] = s->decoded_buffer + FFALIGN(blockstodecode, 8);
+
+ /* get output buffer */
+ frame->nb_samples = blockstodecode;
+ if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) {
+ av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
+ return ret;
+ }
s->error=0;
ape_unpack_stereo(s, blockstodecode);
emms_c();
- if(s->error || s->ptr > s->data_end){
+ if (s->error) {
s->samples=0;
av_log(avctx, AV_LOG_ERROR, "Error decoding frame\n");
return AVERROR_INVALIDDATA;
}
- for (i = 0; i < blockstodecode; i++) {
- *samples++ = s->decoded0[i];
- if(s->channels == 2)
- *samples++ = s->decoded1[i];
+ switch (s->bps) {
+ case 8:
+ for (ch = 0; ch < s->channels; ch++) {
+ sample8 = (uint8_t *)frame->data[ch];
+ for (i = 0; i < blockstodecode; i++)
+ *sample8++ = (s->decoded[ch][i] + 0x80) & 0xff;
+ }
+ break;
+ case 16:
+ for (ch = 0; ch < s->channels; ch++) {
+ sample16 = (int16_t *)frame->data[ch];
+ for (i = 0; i < blockstodecode; i++)
+ *sample16++ = s->decoded[ch][i];
+ }
+ break;
+ case 24:
+ for (ch = 0; ch < s->channels; ch++) {
+ sample24 = (int32_t *)frame->data[ch];
+ for (i = 0; i < blockstodecode; i++)
+ *sample24++ = s->decoded[ch][i] << 8;
+ }
+ break;
}
s->samples -= blockstodecode;
- *data_size = blockstodecode * 2 * s->channels;
- bytes_used = s->samples ? s->ptr - s->last_ptr : buf_size;
- s->last_ptr = s->ptr;
+ *got_frame_ptr = 1;
+
return bytes_used;
}
s->samples= 0;
}
+#define OFFSET(x) offsetof(APEContext, x)
+#define PAR (AV_OPT_FLAG_DECODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM)
+static const AVOption options[] = {
+ { "max_samples", "maximum number of samples decoded per call", OFFSET(blocks_per_loop), AV_OPT_TYPE_INT, { .i64 = 4608 }, 1, INT_MAX, PAR, "max_samples" },
+ { "all", "no maximum. decode all samples for each packet at once", 0, AV_OPT_TYPE_CONST, { .i64 = INT_MAX }, INT_MIN, INT_MAX, PAR, "max_samples" },
+ { NULL},
+};
+
+static const AVClass ape_decoder_class = {
+ .class_name = "APE decoder",
+ .item_name = av_default_item_name,
+ .option = options,
+ .version = LIBAVUTIL_VERSION_INT,
+};
+
AVCodec ff_ape_decoder = {
.name = "ape",
.type = AVMEDIA_TYPE_AUDIO,
- .id = CODEC_ID_APE,
+ .id = AV_CODEC_ID_APE,
.priv_data_size = sizeof(APEContext),
.init = ape_decode_init,
.close = ape_decode_close,
.decode = ape_decode_frame,
- .capabilities = CODEC_CAP_SUBFRAMES,
- .flush = ape_flush,
- .long_name = NULL_IF_CONFIG_SMALL("Monkey's Audio"),
+ .capabilities = CODEC_CAP_SUBFRAMES | CODEC_CAP_DELAY | CODEC_CAP_DR1,
+ .flush = ape_flush,
+ .long_name = NULL_IF_CONFIG_SMALL("Monkey's Audio"),
+ .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_U8P,
+ AV_SAMPLE_FMT_S16P,
+ AV_SAMPLE_FMT_S32P,
+ AV_SAMPLE_FMT_NONE },
+ .priv_class = &ape_decoder_class,
};