* passed through the extradata[_size] fields. This atom is tacked onto
* the end of an 'alac' stsd atom and has the following format:
*
- * 32bit atom size
- * 32bit tag ("alac")
- * 32bit tag version (0)
- * 32bit samples per frame (used when not set explicitly in the frames)
- * 8bit compatible version (0)
- * 8bit sample size
- * 8bit history mult (40)
- * 8bit initial history (14)
- * 8bit rice param limit (10)
- * 8bit channels
- * 16bit maxRun (255)
- * 32bit max coded frame size (0 means unknown)
- * 32bit average bitrate (0 means unknown)
- * 32bit samplerate
+ * 32 bits atom size
+ * 32 bits tag ("alac")
+ * 32 bits tag version (0)
+ * 32 bits samples per frame (used when not set explicitly in the frames)
+ * 8 bits compatible version (0)
+ * 8 bits sample size
+ * 8 bits history mult (40)
+ * 8 bits initial history (14)
+ * 8 bits rice param limit (10)
+ * 8 bits channels
+ * 16 bits maxRun (255)
+ * 32 bits max coded frame size (0 means unknown)
+ * 32 bits average bitrate (0 means unknown)
+ * 32 bits samplerate
*/
-#include "libavutil/audioconvert.h"
+#include <inttypes.h>
+
+#include "libavutil/channel_layout.h"
#include "avcodec.h"
#include "get_bits.h"
#include "bytestream.h"
-#include "unary.h"
+#include "internal.h"
#include "mathops.h"
+#include "unary_legacy.h"
+#include "alac_data.h"
#define ALAC_EXTRADATA_SIZE 36
-#define MAX_CHANNELS 8
-
-typedef struct {
+typedef struct ALACContext {
AVCodecContext *avctx;
- AVFrame frame;
GetBitContext gb;
-
int channels;
- /* buffers */
int32_t *predict_error_buffer[2];
int32_t *output_samples_buffer[2];
int32_t *extra_bits_buffer[2];
uint8_t rice_initial_history;
uint8_t rice_limit;
- int extra_bits; /**< number of extra bits beyond 16-bit */
- int nb_samples; /**< number of samples in the current frame */
+ int extra_bits; /**< number of extra bits beyond 16-bit */
+ int nb_samples; /**< number of samples in the current frame */
} ALACContext;
-enum RawDataBlockType {
- /* At the moment, only SCE, CPE, LFE, and END are recognized. */
- TYPE_SCE,
- TYPE_CPE,
- TYPE_CCE,
- TYPE_LFE,
- TYPE_DSE,
- TYPE_PCE,
- TYPE_FIL,
- TYPE_END
-};
-
-static const uint8_t alac_channel_layout_offsets[8][8] = {
- { 0 },
- { 0, 1 },
- { 2, 0, 1 },
- { 2, 0, 1, 3 },
- { 2, 0, 1, 3, 4 },
- { 2, 0, 1, 4, 5, 3 },
- { 2, 0, 1, 4, 5, 6, 3 },
- { 2, 6, 7, 0, 1, 4, 5, 3 }
-};
-
-static const uint16_t alac_channel_layouts[8] = {
- AV_CH_LAYOUT_MONO,
- AV_CH_LAYOUT_STEREO,
- AV_CH_LAYOUT_SURROUND,
- AV_CH_LAYOUT_4POINT0,
- AV_CH_LAYOUT_5POINT0_BACK,
- AV_CH_LAYOUT_5POINT1_BACK,
- AV_CH_LAYOUT_6POINT1_BACK,
- AV_CH_LAYOUT_7POINT1_WIDE_BACK
-};
-
static inline unsigned int decode_scalar(GetBitContext *gb, int k, int bps)
{
unsigned int x = get_unary_0_9(gb);
int k;
unsigned int x;
- /* read k, that is bits as is */
+ /* calculate rice param and decode next value */
k = av_log2((history >> 9) + 3);
k = FFMIN(k, alac->rice_limit);
x = decode_scalar(&alac->gb, k, bps);
x += sign_modifier;
sign_modifier = 0;
-
output_buffer[i] = (x >> 1) ^ -(x & 1);
- /* now update the history */
+ /* update the history */
if (x > 0xffff)
history = 0xffff;
else
if ((history < 128) && (i + 1 < nb_samples)) {
int block_size;
- k = 7 - av_log2(history) + ((history + 16) >> 6 /* / 64 */);
+ /* calculate rice param and decode block size */
+ k = 7 - av_log2(history) + ((history + 16) >> 6);
k = FFMIN(k, alac->rice_limit);
-
block_size = decode_scalar(&alac->gb, k, 16);
if (block_size > 0) {
block_size * sizeof(*output_buffer));
i += block_size;
}
-
if (block_size <= 0xffff)
sign_modifier = 1;
-
history = 0;
}
}
int lpc_order, int lpc_quant)
{
int i;
+ int32_t *pred = buffer_out;
/* first sample always copies */
*buffer_out = *error_buffer;
}
/* read warm-up samples */
- for (i = 0; i < lpc_order; i++) {
- buffer_out[i + 1] = sign_extend(buffer_out[i] + error_buffer[i + 1],
- bps);
- }
+ for (i = 1; i <= lpc_order; i++)
+ buffer_out[i] = sign_extend(buffer_out[i - 1] + error_buffer[i], bps);
/* NOTE: 4 and 8 are very common cases that could be optimized. */
- /* general case */
- for (i = lpc_order; i < nb_samples - 1; i++) {
+ for (; i < nb_samples; i++) {
int j;
int val = 0;
- int error_val = error_buffer[i + 1];
+ int error_val = error_buffer[i];
int error_sign;
- int d = buffer_out[i - lpc_order];
-
- for (j = 0; j < lpc_order; j++) {
- val += (buffer_out[i - j] - d) * lpc_coefs[j];
- }
+ int d = *pred++;
+ /* LPC prediction */
+ for (j = 0; j < lpc_order; j++)
+ val += (pred[j] - d) * lpc_coefs[j];
val = (val + (1 << (lpc_quant - 1))) >> lpc_quant;
val += d + error_val;
-
- buffer_out[i + 1] = sign_extend(val, bps);
+ buffer_out[i] = sign_extend(val, bps);
/* adapt LPC coefficients */
error_sign = sign_only(error_val);
if (error_sign) {
- for (j = lpc_order - 1; j >= 0 && error_val * error_sign > 0; j--) {
+ for (j = 0; j < lpc_order && error_val * error_sign > 0; j++) {
int sign;
- val = d - buffer_out[i - j];
+ val = d - pred[j];
sign = sign_only(val) * error_sign;
lpc_coefs[j] -= sign;
val *= sign;
- error_val -= (val >> lpc_quant) * (lpc_order - j);
+ error_val -= (val >> lpc_quant) * (j + 1);
}
}
}
}
-static void decorrelate_stereo(int32_t *buffer[2],
- int nb_samples, int decorr_shift,
- int decorr_left_weight)
+static void decorrelate_stereo(int32_t *buffer[2], int nb_samples,
+ int decorr_shift, int decorr_left_weight)
{
int i;
}
}
-static void append_extra_bits(int32_t *buffer[2],
- int32_t *extra_bits_buffer[2],
+static void append_extra_bits(int32_t *buffer[2], int32_t *extra_bits_buffer[2],
int extra_bits, int channels, int nb_samples)
{
int i, ch;
buffer[ch][i] = (buffer[ch][i] << extra_bits) | extra_bits_buffer[ch][i];
}
-static int decode_element(AVCodecContext *avctx, void *data, int ch_index,
+static int decode_element(AVCodecContext *avctx, AVFrame *frame, int ch_index,
int channels)
{
ALACContext *alac = avctx->priv_data;
- int has_size;
- int bps;
- int is_compressed;
- int decorr_shift;
- int decorr_left_weight;
+ int has_size, bps, is_compressed, decorr_shift, decorr_left_weight, ret;
uint32_t output_samples;
- int i, ch, ret;
+ int i, ch;
skip_bits(&alac->gb, 4); /* element instance tag */
skip_bits(&alac->gb, 12); /* unused header bits */
alac->extra_bits = get_bits(&alac->gb, 2) << 3;
bps = alac->sample_size - alac->extra_bits + channels - 1;
if (bps > 32) {
- av_log(avctx, AV_LOG_ERROR, "bps is unsupported: %d\n", bps);
+ avpriv_report_missing_feature(avctx, "bps %d", bps);
return AVERROR_PATCHWELCOME;
}
else
output_samples = alac->max_samples_per_frame;
if (!output_samples || output_samples > alac->max_samples_per_frame) {
- av_log(avctx, AV_LOG_ERROR, "invalid samples per frame: %d\n",
+ av_log(avctx, AV_LOG_ERROR, "invalid samples per frame: %"PRIu32"\n",
output_samples);
return AVERROR_INVALIDDATA;
}
if (!alac->nb_samples) {
/* get output buffer */
- alac->frame.nb_samples = output_samples;
- if ((ret = avctx->get_buffer(avctx, &alac->frame)) < 0) {
+ frame->nb_samples = output_samples;
+ if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) {
av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
return ret;
}
- if (alac->sample_size > 16) {
- for (ch = 0; ch < channels; ch++)
- alac->output_samples_buffer[ch] = (int32_t *)alac->frame.extended_data[ch_index + ch];
- }
} else if (output_samples != alac->nb_samples) {
- av_log(avctx, AV_LOG_ERROR, "sample count mismatch: %u != %d\n",
+ av_log(avctx, AV_LOG_ERROR, "sample count mismatch: %"PRIu32" != %d\n",
output_samples, alac->nb_samples);
return AVERROR_INVALIDDATA;
}
alac->nb_samples = output_samples;
+ if (alac->sample_size > 16) {
+ for (ch = 0; ch < channels; ch++)
+ alac->output_samples_buffer[ch] = (int32_t *)frame->extended_data[ch_index + ch];
+ }
if (is_compressed) {
int16_t lpc_coefs[2][32];
int lpc_quant[2];
int rice_history_mult[2];
+ if (!alac->rice_limit) {
+ avpriv_request_sample(alac->avctx,
+ "Compression with rice limit 0");
+ return AVERROR(ENOSYS);
+ }
+
decorr_shift = get_bits(&alac->gb, 8);
decorr_left_weight = get_bits(&alac->gb, 8);
rice_history_mult[ch] = get_bits(&alac->gb, 3);
lpc_order[ch] = get_bits(&alac->gb, 5);
+ if (lpc_order[ch] >= alac->max_samples_per_frame)
+ return AVERROR_INVALIDDATA;
+
/* read the predictor table */
- for (i = 0; i < lpc_order[ch]; i++)
+ for (i = lpc_order[ch] - 1; i >= 0; i--)
lpc_coefs[ch][i] = get_sbits(&alac->gb, 16);
}
/* not compressed, easy case */
for (i = 0; i < alac->nb_samples; i++) {
for (ch = 0; ch < channels; ch++) {
- alac->output_samples_buffer[ch][i] = get_sbits_long(&alac->gb, alac->sample_size);
+ alac->output_samples_buffer[ch][i] =
+ get_sbits_long(&alac->gb, alac->sample_size);
}
}
- alac->extra_bits = 0;
+ alac->extra_bits = 0;
decorr_shift = 0;
decorr_left_weight = 0;
}
switch(alac->sample_size) {
case 16: {
for (ch = 0; ch < channels; ch++) {
- int16_t *outbuffer = (int16_t *)alac->frame.extended_data[ch_index + ch];
+ int16_t *outbuffer = (int16_t *)frame->extended_data[ch_index + ch];
for (i = 0; i < alac->nb_samples; i++)
*outbuffer++ = alac->output_samples_buffer[ch][i];
}}
int *got_frame_ptr, AVPacket *avpkt)
{
ALACContext *alac = avctx->priv_data;
- enum RawDataBlockType element;
+ AVFrame *frame = data;
+ enum AlacRawDataBlockType element;
int channels;
- int ch, ret;
+ int ch, ret, got_end;
init_get_bits(&alac->gb, avpkt->data, avpkt->size * 8);
+ got_end = 0;
alac->nb_samples = 0;
ch = 0;
- while (get_bits_left(&alac->gb)) {
+ while (get_bits_left(&alac->gb) >= 3) {
element = get_bits(&alac->gb, 3);
- if (element == TYPE_END)
+ if (element == TYPE_END) {
+ got_end = 1;
break;
+ }
if (element > TYPE_CPE && element != TYPE_LFE) {
- av_log(avctx, AV_LOG_ERROR, "syntax element unsupported: %d", element);
+ avpriv_report_missing_feature(avctx, "Syntax element %d", element);
return AVERROR_PATCHWELCOME;
}
channels = (element == TYPE_CPE) ? 2 : 1;
- if (ch + channels > alac->channels) {
+ if (ch + channels > alac->channels ||
+ ff_alac_channel_layout_offsets[alac->channels - 1][ch] + channels > alac->channels) {
av_log(avctx, AV_LOG_ERROR, "invalid element channel count\n");
return AVERROR_INVALIDDATA;
}
- ret = decode_element(avctx, data,
- alac_channel_layout_offsets[alac->channels - 1][ch],
+ ret = decode_element(avctx, frame,
+ ff_alac_channel_layout_offsets[alac->channels - 1][ch],
channels);
- if (ret < 0)
+ if (ret < 0 && get_bits_left(&alac->gb))
return ret;
ch += channels;
}
+ if (!got_end) {
+ av_log(avctx, AV_LOG_ERROR, "no end tag found. incomplete packet.\n");
+ return AVERROR_INVALIDDATA;
+ }
+ if (!alac->nb_samples) {
+ av_log(avctx, AV_LOG_ERROR, "No decodable data in the packet\n");
+ return AVERROR_INVALIDDATA;
+ }
- if (avpkt->size * 8 - get_bits_count(&alac->gb) > 8)
+ if (avpkt->size * 8 - get_bits_count(&alac->gb) > 8) {
av_log(avctx, AV_LOG_ERROR, "Error : %d bits left\n",
avpkt->size * 8 - get_bits_count(&alac->gb));
+ }
- *got_frame_ptr = 1;
- *(AVFrame *)data = alac->frame;
+ *got_frame_ptr = 1;
return avpkt->size;
}
bytestream2_skipu(&gb, 12); // size:4, alac:4, version:4
alac->max_samples_per_frame = bytestream2_get_be32u(&gb);
- if (!alac->max_samples_per_frame || alac->max_samples_per_frame > INT_MAX) {
- av_log(alac->avctx, AV_LOG_ERROR, "max samples per frame invalid: %u\n",
+ if (!alac->max_samples_per_frame ||
+ alac->max_samples_per_frame > INT_MAX / sizeof(int32_t)) {
+ av_log(alac->avctx, AV_LOG_ERROR,
+ "max samples per frame invalid: %"PRIu32"\n",
alac->max_samples_per_frame);
return AVERROR_INVALIDDATA;
}
alac->avctx = avctx;
/* initialize from the extradata */
- if (alac->avctx->extradata_size != ALAC_EXTRADATA_SIZE) {
- av_log(avctx, AV_LOG_ERROR, "alac: expected %d extradata bytes\n",
- ALAC_EXTRADATA_SIZE);
- return -1;
+ if (alac->avctx->extradata_size < ALAC_EXTRADATA_SIZE) {
+ av_log(avctx, AV_LOG_ERROR, "alac: extradata is too small\n");
+ return AVERROR_INVALIDDATA;
}
if (alac_set_info(alac)) {
av_log(avctx, AV_LOG_ERROR, "alac: set_info failed\n");
case 24:
case 32: avctx->sample_fmt = AV_SAMPLE_FMT_S32P;
break;
- default: av_log_ask_for_sample(avctx, "Sample depth %d is not supported.\n",
- alac->sample_size);
+ default: avpriv_request_sample(avctx, "Sample depth %d", alac->sample_size);
return AVERROR_PATCHWELCOME;
}
+ avctx->bits_per_raw_sample = alac->sample_size;
if (alac->channels < 1) {
av_log(avctx, AV_LOG_WARNING, "Invalid channel count\n");
alac->channels = avctx->channels;
} else {
- if (alac->channels > MAX_CHANNELS)
+ if (alac->channels > ALAC_MAX_CHANNELS)
alac->channels = avctx->channels;
else
avctx->channels = alac->channels;
}
- if (avctx->channels > MAX_CHANNELS) {
- av_log(avctx, AV_LOG_ERROR, "Unsupported channel count: %d\n",
- avctx->channels);
+ if (avctx->channels > ALAC_MAX_CHANNELS) {
+ avpriv_report_missing_feature(avctx, "Channel count %d",
+ avctx->channels);
return AVERROR_PATCHWELCOME;
}
- avctx->channel_layout = alac_channel_layouts[alac->channels - 1];
+ avctx->channel_layout = ff_alac_channel_layouts[alac->channels - 1];
if ((ret = allocate_buffers(alac)) < 0) {
av_log(avctx, AV_LOG_ERROR, "Error allocating buffers\n");
return ret;
}
- avcodec_get_frame_defaults(&alac->frame);
- avctx->coded_frame = &alac->frame;
-
return 0;
}
AVCodec ff_alac_decoder = {
.name = "alac",
+ .long_name = NULL_IF_CONFIG_SMALL("ALAC (Apple Lossless Audio Codec)"),
.type = AVMEDIA_TYPE_AUDIO,
- .id = CODEC_ID_ALAC,
+ .id = AV_CODEC_ID_ALAC,
.priv_data_size = sizeof(ALACContext),
.init = alac_decode_init,
.close = alac_decode_close,
.decode = alac_decode_frame,
- .capabilities = CODEC_CAP_DR1,
- .long_name = NULL_IF_CONFIG_SMALL("ALAC (Apple Lossless Audio Codec)"),
+ .capabilities = AV_CODEC_CAP_DR1,
};