diff options
author | Kostya Shishkov <kostya.shishkov@gmail.com> | 2007-09-13 03:22:47 +0000 |
---|---|---|
committer | Kostya Shishkov <kostya.shishkov@gmail.com> | 2007-09-13 03:22:47 +0000 |
commit | bf4a1f17ee9237b6efd4250cf894e274afc31a5f (patch) | |
tree | 0c0e605b516d2e9bb64181daa2ec0819d078eb34 | |
parent | 48fe9238a0aec437aa9ab9a8912191d163feb519 (diff) | |
download | ffmpeg-bf4a1f17ee9237b6efd4250cf894e274afc31a5f.tar.gz |
Monkey Audio decoder
Originally committed as revision 10484 to svn://svn.ffmpeg.org/ffmpeg/trunk
-rw-r--r-- | Changelog | 1 | ||||
-rw-r--r-- | doc/general.texi | 2 | ||||
-rw-r--r-- | libavcodec/Makefile | 1 | ||||
-rw-r--r-- | libavcodec/allcodecs.c | 1 | ||||
-rw-r--r-- | libavcodec/allcodecs.h | 1 | ||||
-rw-r--r-- | libavcodec/apedec.c | 922 | ||||
-rw-r--r-- | libavcodec/avcodec.h | 5 | ||||
-rw-r--r-- | libavformat/Makefile | 1 | ||||
-rw-r--r-- | libavformat/allformats.c | 1 | ||||
-rw-r--r-- | libavformat/allformats.h | 1 | ||||
-rw-r--r-- | libavformat/ape.c | 392 | ||||
-rw-r--r-- | libavformat/avformat.h | 4 |
12 files changed, 1328 insertions, 4 deletions
@@ -94,6 +94,7 @@ version <next> - NUT muxer (since r10052) - Matroska muxer - Slice-based parallel H.264 decoding +- Monkey's Audio demuxer and decoder version 0.4.9-pre1: diff --git a/doc/general.texi b/doc/general.texi index 13c67a5462..542233fbf6 100644 --- a/doc/general.texi +++ b/doc/general.texi @@ -116,6 +116,7 @@ different game cutscenes repacked for use with ScummVM. @tab Used in some games from Bethesda Softworks. @item CRYO APC @tab @tab X @tab Audio format used in some games by CRYO Interactive Entertainment. +@item Monkey's Audio @tab @tab X @end multitable @code{X} means that encoding (resp. decoding) is supported. @@ -311,6 +312,7 @@ following image formats are supported: @tab Only SV7 is supported @item DT$ Coherent Audio @tab @tab X @item ATRAC 3 @tab @tab X +@item Monkey's Audio @tab @tab X @tab Only versions 3.97-3.99 are supported @end multitable @code{X} means that encoding (resp. decoding) is supported. diff --git a/libavcodec/Makefile b/libavcodec/Makefile index 84742d4c06..3efe55295f 100644 --- a/libavcodec/Makefile +++ b/libavcodec/Makefile @@ -35,6 +35,7 @@ OBJS-$(CONFIG_AASC_DECODER) += aasc.o OBJS-$(CONFIG_AC3_DECODER) += ac3dec.o ac3tab.o ac3.o mdct.o fft.o OBJS-$(CONFIG_AC3_ENCODER) += ac3enc.o ac3tab.o ac3.o OBJS-$(CONFIG_ALAC_DECODER) += alac.o +OBJS-$(CONFIG_APE_DECODER) += apedec.o OBJS-$(CONFIG_ASV1_DECODER) += asv1.o OBJS-$(CONFIG_ASV1_ENCODER) += asv1.o OBJS-$(CONFIG_ASV2_DECODER) += asv1.o diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c index 1362451d20..729710474a 100644 --- a/libavcodec/allcodecs.c +++ b/libavcodec/allcodecs.c @@ -168,6 +168,7 @@ void avcodec_register_all(void) REGISTER_DECODER (MPEG4AAC, mpeg4aac); REGISTER_ENCDEC (AC3, ac3); REGISTER_DECODER (ALAC, alac); + REGISTER_DECODER (APE, ape); REGISTER_DECODER (ATRAC3, atrac3); REGISTER_DECODER (COOK, cook); REGISTER_DECODER (DCA, dca); diff --git a/libavcodec/allcodecs.h b/libavcodec/allcodecs.h index 34efcb3bfa..4e94c37ccb 100644 --- a/libavcodec/allcodecs.h +++ b/libavcodec/allcodecs.h @@ -79,6 +79,7 @@ extern AVCodec zmbv_encoder; extern AVCodec aasc_decoder; extern AVCodec ac3_decoder; extern AVCodec alac_decoder; +extern AVCodec ape_decoder; extern AVCodec asv1_decoder; extern AVCodec asv2_decoder; extern AVCodec atrac3_decoder; diff --git a/libavcodec/apedec.c b/libavcodec/apedec.c new file mode 100644 index 0000000000..68358e1868 --- /dev/null +++ b/libavcodec/apedec.c @@ -0,0 +1,922 @@ +/* + * Monkey's Audio lossless audio decoder + * Copyright (c) 2007 Benjamin Zores <ben@geexbox.org> + * based upon libdemac from Dave Chapman. + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * 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 + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#define ALT_BITSTREAM_READER_LE +#include "avcodec.h" +#include "dsputil.h" +#include "bitstream.h" +#include "bytestream.h" + +/** + * @file apedec.c + * Monkey's Audio lossless audio decoder + */ + +#define BLOCKS_PER_LOOP 4608 +#define MAX_CHANNELS 2 +#define MAX_BYTESPERSAMPLE 3 + +#define APE_FRAMECODE_MONO_SILENCE 1 +#define APE_FRAMECODE_STEREO_SILENCE 3 +#define APE_FRAMECODE_PSEUDO_STEREO 4 + +#define HISTORY_SIZE 512 +#define PREDICTOR_ORDER 8 +/** Total size of all predictor histories */ +#define PREDICTOR_SIZE 50 + +#define YDELAYA (18 + PREDICTOR_ORDER*4) +#define YDELAYB (18 + PREDICTOR_ORDER*3) +#define XDELAYA (18 + PREDICTOR_ORDER*2) +#define XDELAYB (18 + PREDICTOR_ORDER) + +#define YADAPTCOEFFSA 18 +#define XADAPTCOEFFSA 14 +#define YADAPTCOEFFSB 10 +#define XADAPTCOEFFSB 5 + +/** + * Possible compression levels + * @{ + */ +enum APECompressionLevel { + COMPRESSION_LEVEL_FAST = 1000, + COMPRESSION_LEVEL_NORMAL = 2000, + COMPRESSION_LEVEL_HIGH = 3000, + COMPRESSION_LEVEL_EXTRA_HIGH = 4000, + COMPRESSION_LEVEL_INSANE = 5000 +}; +/** @} */ + +#define APE_FILTER_LEVELS 3 + +/** Filter orders depending on compression level */ +static const uint16_t ape_filter_orders[5][APE_FILTER_LEVELS] = { + { 0, 0, 0 }, + { 16, 0, 0 }, + { 64, 0, 0 }, + { 32, 256, 0 }, + { 16, 256, 1280 } +}; + +/** Filter fraction bits depending on compression level */ +static const uint16_t ape_filter_fracbits[5][APE_FILTER_LEVELS] = { + { 0, 0, 0 }, + { 11, 0, 0 }, + { 11, 0, 0 }, + { 10, 13, 0 }, + { 11, 13, 15 } +}; + + +/** Filters applied to the decoded data */ +typedef struct APEFilter { + int16_t *coeffs; ///< actual coefficients used in filtering + int16_t *adaptcoeffs; ///< adaptive filter coefficients used for correcting of actual filter coefficients + int16_t *historybuffer; ///< filter memory + int16_t *delay; ///< filtered values + + int avg; +} APEFilter; + +typedef struct APERice { + uint32_t k; + uint32_t ksum; +} APERice; + +typedef struct APERangecoder { + uint32_t low; ///< low end of interval + uint32_t range; ///< length of interval + uint32_t help; ///< bytes_to_follow resp. intermediate value + unsigned int buffer; ///< buffer for input/output +} APERangecoder; + +/** Filter histories */ +typedef struct APEPredictor { + int32_t *buf; + + int32_t lastA[2]; + + int32_t filterA[2]; + int32_t filterB[2]; + + int32_t coeffsA[2][4]; ///< adaption coefficients + int32_t coeffsB[2][5]; ///< adaption coefficients + int32_t historybuffer[HISTORY_SIZE + PREDICTOR_SIZE]; +} APEPredictor; + +/** Decoder context */ +typedef struct APEContext { + AVCodecContext *avctx; + DSPContext dsp; + int channels; + int samples; ///< samples left to decode in current frame + + int fileversion; ///< codec version, very important in decoding process + int compression_level; ///< compression levels + int fset; ///< which filter set to use (calculated from compression level) + int flags; ///< global decoder flags + + 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 + + int16_t* filterbuf[APE_FILTER_LEVELS]; ///< filter memory + + APERangecoder rc; ///< rangecoder used to decode actual values + APERice riceX; ///< rice code parameters for the second channel + APERice riceY; ///< rice code parameters for the first channel + APEFilter filters[APE_FILTER_LEVELS][2]; ///< filters used for reconstruction + + uint8_t *data; ///< current frame data + uint8_t *data_end; ///< frame data end + uint8_t *ptr; ///< current position in frame data + uint8_t *last_ptr; ///< position where last 4608-sample block ended +} APEContext; + +// TODO: dsputilize +static inline void vector_add(int16_t * v1, int16_t * v2, int order) +{ + while (order--) + *v1++ += *v2++; +} + +// TODO: dsputilize +static inline void vector_sub(int16_t * v1, int16_t * v2, int order) +{ + while (order--) + *v1++ -= *v2++; +} + +// TODO: dsputilize +static inline int32_t scalarproduct(int16_t * v1, int16_t * v2, int order) +{ + int res = 0; + + while (order--) + res += *v1++ * *v2++; + + return res; +} + +static int ape_decode_init(AVCodecContext * avctx) +{ + APEContext *s = avctx->priv_data; + int i; + + if (avctx->extradata_size != 6) { + av_log(avctx, AV_LOG_ERROR, "Incorrect extradata\n"); + return -1; + } + if (avctx->bits_per_sample != 16) { + av_log(avctx, AV_LOG_ERROR, "Only 16-bit samples are supported\n"); + return -1; + } + if (avctx->channels > 2) { + av_log(avctx, AV_LOG_ERROR, "Only mono and stereo is supported\n"); + return -1; + } + s->avctx = avctx; + s->channels = avctx->channels; + s->fileversion = AV_RL16(avctx->extradata); + s->compression_level = AV_RL16(avctx->extradata + 2); + s->flags = AV_RL16(avctx->extradata + 4); + + av_log(avctx, AV_LOG_DEBUG, "Compression Level: %d - Flags: %d\n", s->compression_level, s->flags); + if (s->compression_level % 1000 || s->compression_level > COMPRESSION_LEVEL_INSANE) { + av_log(avctx, AV_LOG_ERROR, "Incorrect compression level %d\n", s->compression_level); + return -1; + } + s->fset = s->compression_level / 1000 - 1; + for (i = 0; i < APE_FILTER_LEVELS; i++) { + if (!ape_filter_orders[s->fset][i]) + break; + s->filterbuf[i] = av_malloc((ape_filter_orders[s->fset][i] * 3 + HISTORY_SIZE) * 4); + } + + dsputil_init(&s->dsp, avctx); + return 0; +} + +static int ape_decode_close(AVCodecContext * avctx) +{ + APEContext *s = avctx->priv_data; + int i; + + for (i = 0; i < APE_FILTER_LEVELS; i++) + av_freep(&s->filterbuf[i]); + + return 0; +} + +/** + * @defgroup rangecoder APE range decoder + * @{ + */ + +#define CODE_BITS 32 +#define TOP_VALUE ((unsigned int)1 << (CODE_BITS-1)) +#define SHIFT_BITS (CODE_BITS - 9) +#define EXTRA_BITS ((CODE_BITS-2) % 8 + 1) +#define BOTTOM_VALUE (TOP_VALUE >> 8) + +/** Start the decoder */ +static inline void range_start_decoding(APEContext * ctx) +{ + ctx->rc.buffer = bytestream_get_byte(&ctx->ptr); + ctx->rc.low = ctx->rc.buffer >> (8 - EXTRA_BITS); + ctx->rc.range = (uint32_t) 1 << EXTRA_BITS; +} + +/** Perform normalization */ +static inline void range_dec_normalize(APEContext * ctx) +{ + while (ctx->rc.range <= BOTTOM_VALUE) { + ctx->rc.buffer = (ctx->rc.buffer << 8) | bytestream_get_byte(&ctx->ptr); + ctx->rc.low = (ctx->rc.low << 8) | ((ctx->rc.buffer >> 1) & 0xFF); + ctx->rc.range <<= 8; + } +} + +/** + * Calculate culmulative frequency for next symbol. Does NO update! + * @param tot_f is the total frequency or (code_value)1<<shift + * @return the culmulative frequency + */ +static inline int range_decode_culfreq(APEContext * ctx, int tot_f) +{ + range_dec_normalize(ctx); + ctx->rc.help = ctx->rc.range / tot_f; + return ctx->rc.low / ctx->rc.help; +} + +/** + * Decode value with given size in bits + * @param shift number of bits to decode + */ +static inline int range_decode_culshift(APEContext * ctx, int shift) +{ + range_dec_normalize(ctx); + ctx->rc.help = ctx->rc.range >> shift; + return ctx->rc.low / ctx->rc.help; +} + + +/** + * Update decoding state + * @param sy_f the interval length (frequency of the symbol) + * @param lt_f the lower end (frequency sum of < symbols) + */ +static inline void range_decode_update(APEContext * ctx, int sy_f, int lt_f) +{ + ctx->rc.low -= ctx->rc.help * lt_f; + ctx->rc.range = ctx->rc.help * sy_f; +} + +/** Decode n bits (n <= 16) without modelling */ +static inline int range_decode_bits(APEContext * ctx, int n) +{ + int sym = range_decode_culshift(ctx, n); + range_decode_update(ctx, 1, sym); + return sym; +} + + +#define MODEL_ELEMENTS 64 + +/** + * Fixed probabilities for symbols in Monkey Audio version 3.97 + */ +static const uint32_t counts_3970[65] = { + 0, 14824, 28224, 39348, 47855, 53994, 58171, 60926, + 62682, 63786, 64463, 64878, 65126, 65276, 65365, 65419, + 65450, 65469, 65480, 65487, 65491, 65493, 65494, 65495, + 65496, 65497, 65498, 65499, 65500, 65501, 65502, 65503, + 65504, 65505, 65506, 65507, 65508, 65509, 65510, 65511, + 65512, 65513, 65514, 65515, 65516, 65517, 65518, 65519, + 65520, 65521, 65522, 65523, 65524, 65525, 65526, 65527, + 65528, 65529, 65530, 65531, 65532, 65533, 65534, 65535, + 65536 +}; + +/** + * Probability ranges for symbols in Monkey Audio version 3.97 + */ +static const uint16_t counts_diff_3970[64] = { + 14824, 13400, 11124, 8507, 6139, 4177, 2755, 1756, + 1104, 677, 415, 248, 150, 89, 54, 31, + 19, 11, 7, 4, 2, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1 +}; + +/** + * Fixed probabilities for symbols in Monkey Audio version 3.98 + */ +static const uint32_t counts_3980[65] = { + 0, 19578, 36160, 48417, 56323, 60899, 63265, 64435, + 64971, 65232, 65351, 65416, 65447, 65466, 65476, 65482, + 65485, 65488, 65490, 65491, 65492, 65493, 65494, 65495, + 65496, 65497, 65498, 65499, 65500, 65501, 65502, 65503, + 65504, 65505, 65506, 65507, 65508, 65509, 65510, 65511, + 65512, 65513, 65514, 65515, 65516, 65517, 65518, 65519, + 65520, 65521, 65522, 65523, 65524, 65525, 65526, 65527, + 65528, 65529, 65530, 65531, 65532, 65533, 65534, 65535, + 65536 +}; + +/** + * Probability ranges for symbols in Monkey Audio version 3.98 + */ +static const uint16_t counts_diff_3980[64] = { + 19578, 16582, 12257, 7906, 4576, 2366, 1170, 536, + 261, 119, 65, 31, 19, 10, 6, 3, + 3, 2, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1 +}; + +/** + * Decode symbol + * @param counts probability range start position + * @param count_diffs probability range widths + */ +static inline int range_get_symbol(APEContext * ctx, + const uint32_t counts[], + const uint16_t counts_diff[]) +{ + int symbol, cf; + + cf = range_decode_culshift(ctx, 16); + + /* figure out the symbol inefficiently; a binary search would be much better */ + for (symbol = 0; counts[symbol + 1] <= cf; symbol++); + + range_decode_update(ctx, counts_diff[symbol], counts[symbol]); + + return symbol; +} +/** @} */ // group rangecoder + +static inline void update_rice(APERice *rice, int x) +{ + rice->ksum += ((x + 1) / 2) - ((rice->ksum + 16) >> 5); + + if (rice->k == 0) + rice->k = 1; + else if (rice->ksum < (1 << (rice->k + 4))) + rice->k--; + else if (rice->ksum >= (1 << (rice->k + 5))) + rice->k++; +} + +static inline int ape_decode_value(APEContext * ctx, APERice *rice) +{ + int x, overflow; + + if (ctx->fileversion < 3980) { + int tmpk; + + overflow = range_get_symbol(ctx, counts_3970, counts_diff_3970); + + if (overflow == (MODEL_ELEMENTS - 1)) { + tmpk = range_decode_bits(ctx, 5); + overflow = 0; + } else + tmpk = (rice->k < 1) ? 0 : rice->k - 1; + + if (tmpk <= 16) + x = range_decode_bits(ctx, tmpk); + else { + x = range_decode_bits(ctx, 16); + x |= (range_decode_bits(ctx, tmpk - 16) << 16); + } + x += overflow << tmpk; + } else { + int base, pivot; + + pivot = rice->ksum >> 5; + if (pivot == 0) + pivot = 1; + + overflow = range_get_symbol(ctx, counts_3980, counts_diff_3980); + + if (overflow == (MODEL_ELEMENTS - 1)) { + overflow = range_decode_bits(ctx, 16) << 16; + overflow |= range_decode_bits(ctx, 16); + } + + base = range_decode_culfreq(ctx, pivot); + range_decode_update(ctx, 1, base); + + x = base + overflow * pivot; + } + + update_rice(rice, x); + + /* Convert to signed */ + if (x & 1) + return (x >> 1) + 1; + else + return -(x >> 1); +} + +static void entropy_decode(APEContext * ctx, int blockstodecode, int stereo) +{ + int32_t *decoded0 = ctx->decoded0; + int32_t *decoded1 = ctx->decoded1; + + 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); + } + } + + if (ctx->blocksdecoded == ctx->currentframeblocks) + range_dec_normalize(ctx); /* normalize to use up all bytes */ +} + +static void init_entropy_decoder(APEContext * ctx) +{ + /* Read the CRC */ + ctx->CRC = bytestream_get_be32(&ctx->ptr); + + /* Read the frame flags if they exist */ + ctx->frameflags = 0; + if ((ctx->fileversion > 3820) && (ctx->CRC & 0x80000000)) { + ctx->CRC &= ~0x80000000; + + ctx->frameflags = bytestream_get_be32(&ctx->ptr); + } + + /* Keep a count of the blocks decoded in this frame */ + ctx->blocksdecoded = 0; + + /* Initialise the rice structs */ + ctx->riceX.k = 10; + ctx->riceX.ksum = (1 << ctx->riceX.k) * 16; + ctx->riceY.k = 10; + ctx->riceY.ksum = (1 << ctx->riceY.k) * 16; + + /* The first 8 bits of input are ignored. */ + ctx->ptr++; + + range_start_decoding(ctx); +} + +static const int32_t initial_coeffs[4] = { + 360, 317, -109, 98 +}; + +static void init_predictor_decoder(APEContext * ctx) +{ + APEPredictor *p = &ctx->predictor; + + /* Zero the history buffers */ + memset(p->historybuffer, 0, PREDICTOR_SIZE * sizeof(int32_t)); + p->buf = p->historybuffer; + + /* Initialise and zero the co-efficients */ + memcpy(p->coeffsA[0], initial_coeffs, sizeof(initial_coeffs)); + memcpy(p->coeffsA[1], initial_coeffs, sizeof(initial_coeffs)); + memset(p->coeffsB, 0, sizeof(p->coeffsB)); + + p->filterA[0] = p->filterA[1] = 0; + p->filterB[0] = p->filterB[1] = 0; + p->lastA[0] = p->lastA[1] = 0; +} + +/** Get inverse sign of integer (-1 for positive, 1 for negative and 0 for zero) */ +static inline int APESIGN(int32_t x) { + return (x < 0) - (x > 0); +} + +static int predictor_update_filter(APEPredictor *p, const int decoded, const int filter, const int delayA, const int delayB, const int adaptA, const int adaptB) +{ + int32_t predictionA, predictionB; + + p->buf[delayA] = p->lastA[filter]; + p->buf[adaptA] = APESIGN(p->buf[delayA]); + p->buf[delayA - 1] = p->buf[delayA] - p->buf[delayA - 1]; + p->buf[adaptA - 1] = APESIGN(p->buf[delayA - 1]); + + predictionA = p->buf[delayA ] * p->coeffsA[filter][0] + + p->buf[delayA - 1] * p->coeffsA[filter][1] + + p->buf[delayA - 2] * p->coeffsA[filter][2] + + p->buf[delayA - 3] * p->coeffsA[filter][3]; + + /* Apply a scaled first-order filter compression */ + p->buf[delayB] = p->filterA[filter ^ 1] - ((p->filterB[filter] * 31) >> 5); + p->buf[adaptB] = APESIGN(p->buf[delayB]); + p->buf[delayB - 1] = p->buf[delayB] - p->buf[delayB - 1]; + p->buf[adaptB - 1] = APESIGN(p->buf[delayB - 1]); + p->filterB[filter] = p->filterA[filter ^ 1]; + + predictionB = p->buf[delayB ] * p->coeffsB[filter][0] + + p->buf[delayB - 1] * p->coeffsB[filter][1] + + p->buf[delayB - 2] * p->coeffsB[filter][2] + + p->buf[delayB - 3] * p->coeffsB[filter][3] + + p->buf[delayB - 4] * p->coeffsB[filter][4]; + + p->lastA[filter] = decoded + ((predictionA + (predictionB >> 1)) >> 10); + p->filterA[filter] = p->lastA[filter] + ((p->filterA[filter] * 31) >> 5); + + if (!decoded) // no need updating filter coefficients + return p->filterA[filter]; + + if (decoded > 0) { + p->coeffsA[filter][0] -= p->buf[adaptA ]; + p->coeffsA[filter][1] -= p->buf[adaptA - 1]; + p->coeffsA[filter][2] -= p->buf[adaptA - 2]; + p->coeffsA[filter][3] -= p->buf[adaptA - 3]; + + p->coeffsB[filter][0] -= p->buf[adaptB ]; + p->coeffsB[filter][1] -= p->buf[adaptB - 1]; + p->coeffsB[filter][2] -= p->buf[adaptB - 2]; + p->coeffsB[filter][3] -= p->buf[adaptB - 3]; + p->coeffsB[filter][4] -= p->buf[adaptB - 4]; + } else { + p->coeffsA[filter][0] += p->buf[adaptA ]; + p->coeffsA[filter][1] += p->buf[adaptA - 1]; + p->coeffsA[filter][2] += p->buf[adaptA - 2]; + p->coeffsA[filter][3] += p->buf[adaptA - 3]; + + p->coeffsB[filter][0] += p->buf[adaptB ]; + p->coeffsB[filter][1] += p->buf[adaptB - 1]; + p->coeffsB[filter][2] += p->buf[adaptB - 2]; + p->coeffsB[filter][3] += p->buf[adaptB - 3]; + p->coeffsB[filter][4] += p->buf[adaptB - 4]; + } + return p->filterA[filter]; +} + +static void predictor_decode_stereo(APEContext * ctx, int count) +{ + int32_t predictionA, predictionB; + APEPredictor *p = &ctx->predictor; + int32_t *decoded0 = ctx->decoded0; + int32_t *decoded1 = ctx->decoded1; + + while (count--) { + /* Predictor Y */ + predictionA = predictor_update_filter(p, *decoded0, 0, YDELAYA, YDELAYB, YADAPTCOEFFSA, YADAPTCOEFFSB); + predictionB = predictor_update_filter(p, *decoded1, 1, XDELAYA, XDELAYB, XADAPTCOEFFSA, XADAPTCOEFFSB); + *(decoded0++) = predictionA; + *(decoded1++) = predictionB; + + /* Combined */ + p->buf++; + + /* Have we filled the history buffer? */ + if (p->buf == p->historybuffer + HISTORY_SIZE) { + memmove(p->historybuffer, p->buf, PREDICTOR_SIZE * sizeof(int32_t)); + p->buf = p->historybuffer; + } + } +} + +static void predictor_decode_mono(APEContext * ctx, int count) +{ + APEPredictor *p = &ctx->predictor; + int32_t *decoded0 = ctx->decoded0; + int32_t predictionA, currentA, A; + + currentA = p->lastA[0]; + + while (count--) { + A = *decoded0; + + p->buf[YDELAYA] = currentA; + p->buf[YDELAYA - 1] = p->buf[YDELAYA] - p->buf[YDELAYA - 1]; + + predictionA = p->buf[YDELAYA ] * p->coeffsA[0][0] + + p->buf[YDELAYA - 1] * p->coeffsA[0][1] + + p->buf[YDELAYA - 2] * p->coeffsA[0][2] + + p->buf[YDELAYA - 3] * p->coeffsA[0][3]; + + currentA = A + (predictionA >> 10); + + p->buf[YADAPTCOEFFSA] = APESIGN(p->buf[YDELAYA ]); + p->buf[YADAPTCOEFFSA - 1] = APESIGN(p->buf[YDELAYA - 1]); + + if (A > 0) { + p->coeffsA[0][0] -= p->buf[YADAPTCOEFFSA ]; + p->coeffsA[0][1] -= p->buf[YADAPTCOEFFSA - 1]; + p->coeffsA[0][2] -= p->buf[YADAPTCOEFFSA - 2]; + p->coeffsA[0][3] -= p->buf[YADAPTCOEFFSA - 3]; + } else if (A < 0) { + p->coeffsA[0][0] += p->buf[YADAPTCOEFFSA ]; + p->coeffsA[0][1] += p->buf[YADAPTCOEFFSA - 1]; + p->coeffsA[0][2] += p->buf[YADAPTCOEFFSA - 2]; + p->coeffsA[0][3] += p->buf[YADAPTCOEFFSA - 3]; + } + + p->buf++; + + /* Have we filled the history buffer? */ + if (p->buf == p->historybuffer + HISTORY_SIZE) { + memmove(p->historybuffer, p->buf, PREDICTOR_SIZE * sizeof(int32_t)); + p->buf = p->historybuffer; + } + + p->filterA[0] = currentA + ((p->filterA[0] * 31) >> 5); + *(decoded0++) = p->filterA[0]; + } + + p->lastA[0] = currentA; +} + +static void do_init_filter(APEFilter *f, int16_t * buf, int order) +{ + f->coeffs = buf; + f->historybuffer = buf + order; + 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)); + f->avg = 0; +} + +static void init_filter(APEContext * ctx, APEFilter *f, int16_t * buf, int order) +{ + do_init_filter(&f[0], buf, order); + do_init_filter(&f[1], buf + order * 3 + HISTORY_SIZE, order); +} + +static inline void do_apply_filter(int version, APEFilter *f, int32_t *data, int count, int order, int fracbits) +{ + int res; + int absres; + + while (count--) { + /* round fixedpoint scalar product */ + res = (scalarproduct(f->delay - order, f->coeffs, order) + (1 << (fracbits - 1))) >> fracbits; + + if (*data < 0) + vector_add(f->coeffs, f->adaptcoeffs - order, order); + else if (*data > 0) + vector_sub(f->coeffs, f->adaptcoeffs - order, order); + + res += *data; + + *data++ = res; + + /* Update the output history */ + *f->delay++ = av_clip_int16(res); + + if (version < 3980) { + /* Version ??? to < 3.98 files (untested) */ + f->adaptcoeffs[0] = (res == 0) ? 0 : ((res >> 28) & 8) - 4; + f->adaptcoeffs[-4] >>= 1; + f->adaptcoeffs[-8] >>= 1; + } else { + /* Version 3.98 and later files */ + + /* Update the adaption coefficients */ + absres = (res < 0 ? -res : res); + + if (absres > (f->avg * 3)) + *f->adaptcoeffs = ((res >> 25) & 64) - 32; + else if (absres > (f->avg * 4) / 3) + *f->adaptcoeffs = ((res >> 26) & 32) - 16; + else if (absres > 0) + *f->adaptcoeffs = ((res >> 27) & 16) - 8; + else + *f->adaptcoeffs = 0; + + f->avg += (absres - f->avg) / 16; + + f->adaptcoeffs[-1] >>= 1; + f->adaptcoeffs[-2] >>= 1; + f->adaptcoeffs[-8] >>= 1; + } + + f->adaptcoeffs++; + + /* 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)); + f->delay = f->historybuffer + order * 2; + f->adaptcoeffs = f->historybuffer + order; + } + } +} + +static void apply_filter(APEContext * ctx, APEFilter *f, + int32_t * data0, int32_t * data1, + int count, int order, int fracbits) +{ + do_apply_filter(ctx->fileversion, &f[0], data0, count, order, fracbits); + if (data1) + do_apply_filter(ctx->fileversion, &f[1], data1, count, order, fracbits); +} + +static void ape_apply_filters(APEContext * ctx, int32_t * decoded0, + int32_t * decoded1, int count) +{ + int i; + + for (i = 0; i < APE_FILTER_LEVELS; i++) { + if (!ape_filter_orders[ctx->fset][i]) + break; + apply_filter(ctx, ctx->filters[i], decoded0, decoded1, count, ape_filter_orders[ctx->fset][i], ape_filter_fracbits[ctx->fset][i]); + } +} + +static void init_frame_decoder(APEContext * ctx) +{ + int i; + init_entropy_decoder(ctx); + init_predictor_decoder(ctx); + + for (i = 0; i < APE_FILTER_LEVELS; i++) { + if (!ape_filter_orders[ctx->fset][i]) + break; + init_filter(ctx, ctx->filters[i], ctx->filterbuf[i], ape_filter_orders[ctx->fset][i]); + } +} + +static void ape_unpack_mono(APEContext * ctx, int count) +{ + int32_t left; + 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); + + /* Now apply the predictor decoding */ + predictor_decode_mono(ctx, count); + + /* Pseudo-stereo - just copy left channel to right channel */ + if (ctx->channels == 2) { + while (count--) { + left = *decoded0; + *(decoded1++) = *(decoded0++) = left; + } + } +} + +static void ape_unpack_stereo(APEContext * ctx, int count) +{ + int32_t left, right; + int32_t *decoded0 = ctx->decoded0; + int32_t *decoded1 = ctx->decoded1; + + if (ctx->frameflags & APE_FRAMECODE_STEREO_SILENCE) { + /* We are pure silence, so we're done. */ + av_log(ctx->avctx, AV_LOG_DEBUG, "pure silence stereo\n"); + return; + } + + entropy_decode(ctx, count, 1); + ape_apply_filters(ctx, decoded0, decoded1, count); + + /* Now apply the predictor decoding */ + predictor_decode_stereo(ctx, count); + + /* Decorrelate and scale to output depth */ + while (count--) { + left = *decoded1 - (*decoded0 / 2); + right = left + *decoded0; + + *(decoded0++) = left; + *(decoded1++) = right; + } +} + +static int ape_decode_frame(AVCodecContext * avctx, + void *data, int *data_size, + uint8_t * buf, int buf_size) +{ + APEContext *s = avctx->priv_data; + int16_t *samples = data; + int nblocks; + int i, n; + int blockstodecode; + int bytes_used; + + if (buf_size == 0 && !s->samples) { + *data_size = 0; + return 0; + } + + /* should not happen but who knows */ + if (BLOCKS_PER_LOOP * 2 * avctx->channels > *data_size) { + av_log (avctx, AV_LOG_ERROR, "Packet size is too big to be handled in lavc! (max is %d where you have %d)\n", *data_size, s->samples * 2 * avctx->channels); + return -1; + } + + if(!s->samples){ + s->data = av_realloc(s->data, (buf_size + 3) & ~3); + s->dsp.bswap_buf(s->data, buf, buf_size >> 2); + s->ptr = s->last_ptr = s->data; + s->data_end = s->data + buf_size; + + nblocks = s->samples = bytestream_get_be32(&s->ptr); + n = bytestream_get_be32(&s->ptr); + if(n < 0 || n > 3){ + av_log(avctx, AV_LOG_ERROR, "Incorrect offset passed\n"); + s->data = NULL; + return -1; + } + s->ptr += n; + + s->currentframeblocks = nblocks; + buf += 4; + if (s->samples <= 0) { + *data_size = 0; + return buf_size; + } + + memset(s->decoded0, 0, sizeof(s->decoded0)); + memset(s->decoded1, 0, sizeof(s->decoded1)); + + /* Initialize the frame decoder */ + init_frame_decoder(s); + } + + if (!s->data) { + *data_size = 0; + return buf_size; + } + + nblocks = s->samples; + blockstodecode = FFMIN(BLOCKS_PER_LOOP, nblocks); + + if ((s->channels == 1) || (s->frameflags & APE_FRAMECODE_PSEUDO_STEREO)) + ape_unpack_mono(s, blockstodecode); + else + ape_unpack_stereo(s, blockstodecode); + + for (i = 0; i < blockstodecode; i++) { + *samples++ = s->decoded0[i]; + if(s->channels == 2) + *samples++ = s->decoded1[i]; + } + + 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; + return bytes_used; +} + +AVCodec ape_decoder = { + "ape", + CODEC_TYPE_AUDIO, + CODEC_ID_APE, + sizeof(APEContext), + ape_decode_init, + NULL, + ape_decode_close, + ape_decode_frame, +}; diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h index 53f24f25d0..5ec822a2cb 100644 --- a/libavcodec/avcodec.h +++ b/libavcodec/avcodec.h @@ -33,8 +33,8 @@ #define AV_STRINGIFY(s) AV_TOSTRING(s) #define AV_TOSTRING(s) #s -#define LIBAVCODEC_VERSION_INT ((51<<16)+(43<<8)+0) -#define LIBAVCODEC_VERSION 51.43.0 +#define LIBAVCODEC_VERSION_INT ((51<<16)+(44<<8)+0) +#define LIBAVCODEC_VERSION 51.44.0 #define LIBAVCODEC_BUILD LIBAVCODEC_VERSION_INT #define LIBAVCODEC_IDENT "Lavc" AV_STRINGIFY(LIBAVCODEC_VERSION) @@ -260,6 +260,7 @@ enum CodecID { CODEC_ID_GSM_MS, /* as found in WAV */ CODEC_ID_ATRAC3, CODEC_ID_VOXWARE, + CODEC_ID_APE, /* subtitle codecs */ CODEC_ID_DVD_SUBTITLE= 0x17000, diff --git a/libavformat/Makefile b/libavformat/Makefile index c7d4fadd88..1023199413 100644 --- a/libavformat/Makefile +++ b/libavformat/Makefile @@ -20,6 +20,7 @@ OBJS-$(CONFIG_AIFF_MUXER) += aiff.o riff.o OBJS-$(CONFIG_AMR_DEMUXER) += amr.o OBJS-$(CONFIG_AMR_MUXER) += amr.o OBJS-$(CONFIG_APC_DEMUXER) += apc.o +OBJS-$(CONFIG_APE_DEMUXER) += ape.o OBJS-$(CONFIG_ASF_DEMUXER) += asf.o riff.o OBJS-$(CONFIG_ASF_MUXER) += asf-enc.o riff.o OBJS-$(CONFIG_ASF_STREAM_MUXER) += asf-enc.o riff.o diff --git a/libavformat/allformats.c b/libavformat/allformats.c index 0a783ba6c4..1b7af2aca2 100644 --- a/libavformat/allformats.c +++ b/libavformat/allformats.c @@ -53,6 +53,7 @@ void av_register_all(void) REGISTER_MUXDEMUX (AIFF, aiff); REGISTER_MUXDEMUX (AMR, amr); REGISTER_DEMUXER (APC, apc); + REGISTER_DEMUXER (APE, ape); REGISTER_MUXDEMUX (ASF, asf); REGISTER_MUXER (ASF_STREAM, asf_stream); REGISTER_MUXDEMUX (AU, au); diff --git a/libavformat/allformats.h b/libavformat/allformats.h index eb51c9620c..ff66203816 100644 --- a/libavformat/allformats.h +++ b/libavformat/allformats.h @@ -29,6 +29,7 @@ extern AVInputFormat ac3_demuxer; extern AVInputFormat aiff_demuxer; extern AVInputFormat amr_demuxer; extern AVInputFormat apc_demuxer; +extern AVInputFormat ape_demuxer; extern AVInputFormat asf_demuxer; extern AVInputFormat au_demuxer; extern AVInputFormat audio_beos_demuxer; diff --git a/libavformat/ape.c b/libavformat/ape.c new file mode 100644 index 0000000000..336871484e --- /dev/null +++ b/libavformat/ape.c @@ -0,0 +1,392 @@ +/* + * Monkey's Audio APE demuxer + * Copyright (c) 2007 Benjamin Zores <ben@geexbox.org> + * based upon libdemac from Dave Chapman. + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * 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 + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdio.h> + +#include "avformat.h" + +/* The earliest and latest file formats supported by this library */ +#define APE_MIN_VERSION 3970 +#define APE_MAX_VERSION 3990 + +#define MAC_FORMAT_FLAG_8_BIT 1 // is 8-bit [OBSOLETE] +#define MAC_FORMAT_FLAG_CRC 2 // uses the new CRC32 error detection [OBSOLETE] +#define MAC_FORMAT_FLAG_HAS_PEAK_LEVEL 4 // uint32 nPeakLevel after the header [OBSOLETE] +#define MAC_FORMAT_FLAG_24_BIT 8 // is 24-bit [OBSOLETE] +#define MAC_FORMAT_FLAG_HAS_SEEK_ELEMENTS 16 // has the number of seek elements after the peak level +#define MAC_FORMAT_FLAG_CREATE_WAV_HEADER 32 // create the wave header on decompression (not stored) + +#define MAC_SUBFRAME_SIZE 4608 + +#define APE_EXTRADATA_SIZE 6 + +typedef struct { + int64_t pos; + int nblocks; + int size; + int skip; + int64_t pts; +} APEFrame; + +typedef struct { + /* Derived fields */ + uint32_t junklength; + uint32_t firstframe; + uint32_t totalsamples; + int currentframe; + APEFrame *frames; + + /* Info from Descriptor Block */ + char magic[4]; + int16_t fileversion; + int16_t padding1; + uint32_t descriptorlength; + uint32_t headerlength; + uint32_t seektablelength; + uint32_t wavheaderlength; + uint32_t audiodatalength; + uint32_t audiodatalength_high; + uint32_t wavtaillength; + uint8_t md5[16]; + + /* Info from Header Block */ + uint16_t compressiontype; + uint16_t formatflags; + uint32_t blocksperframe; + uint32_t finalframeblocks; + uint32_t totalframes; + uint16_t bps; + uint16_t channels; + uint32_t samplerate; + + /* Seektable */ + uint32_t *seektable; +} APEContext; + +static int ape_probe(AVProbeData * p) +{ + if (p->buf[0] == 'M' && p->buf[1] == 'A' && p->buf[2] == 'C' && p->buf[3] == ' ') + return AVPROBE_SCORE_MAX; + + return 0; +} + +static void ape_dumpinfo(APEContext * ape_ctx) +{ + int i; + + av_log(NULL, AV_LOG_DEBUG, "Descriptor Block:\n\n"); + av_log(NULL, AV_LOG_DEBUG, "magic = \"%c%c%c%c\"\n", ape_ctx->magic[0], ape_ctx->magic[1], ape_ctx->magic[2], ape_ctx->magic[3]); + av_log(NULL, AV_LOG_DEBUG, "fileversion = %d\n", ape_ctx->fileversion); + av_log(NULL, AV_LOG_DEBUG, "descriptorlength = %d\n", ape_ctx->descriptorlength); + av_log(NULL, AV_LOG_DEBUG, "headerlength = %d\n", ape_ctx->headerlength); + av_log(NULL, AV_LOG_DEBUG, "seektablelength = %d\n", ape_ctx->seektablelength); + av_log(NULL, AV_LOG_DEBUG, "wavheaderlength = %d\n", ape_ctx->wavheaderlength); + av_log(NULL, AV_LOG_DEBUG, "audiodatalength = %d\n", ape_ctx->audiodatalength); + av_log(NULL, AV_LOG_DEBUG, "audiodatalength_high = %d\n", ape_ctx->audiodatalength_high); + av_log(NULL, AV_LOG_DEBUG, "wavtaillength = %d\n", ape_ctx->wavtaillength); + av_log(NULL, AV_LOG_DEBUG, "md5 = "); + for (i = 0; i < 16; i++) + av_log(NULL, AV_LOG_DEBUG, "%02x", ape_ctx->md5[i]); + av_log(NULL, AV_LOG_DEBUG, "\n"); + + av_log(NULL, AV_LOG_DEBUG, "\nHeader Block:\n\n"); + + av_log(NULL, AV_LOG_DEBUG, "compressiontype = %d\n", ape_ctx->compressiontype); + av_log(NULL, AV_LOG_DEBUG, "formatflags = %d\n", ape_ctx->formatflags); + av_log(NULL, AV_LOG_DEBUG, "blocksperframe = %d\n", ape_ctx->blocksperframe); + av_log(NULL, AV_LOG_DEBUG, "finalframeblocks = %d\n", ape_ctx->finalframeblocks); + av_log(NULL, AV_LOG_DEBUG, "totalframes = %d\n", ape_ctx->totalframes); + av_log(NULL, AV_LOG_DEBUG, "bps = %d\n", ape_ctx->bps); + av_log(NULL, AV_LOG_DEBUG, "channels = %d\n", ape_ctx->channels); + av_log(NULL, AV_LOG_DEBUG, "samplerate = %d\n", ape_ctx->samplerate); + + av_log(NULL, AV_LOG_DEBUG, "\nSeektable\n\n"); + if ((ape_ctx->seektablelength / sizeof(uint32_t)) != ape_ctx->totalframes) { + av_log(NULL, AV_LOG_DEBUG, "No seektable\n"); + } else { + for (i = 0; i < ape_ctx->seektablelength / sizeof(uint32_t); i++) { + if (i < ape_ctx->totalframes - 1) { + av_log(NULL, AV_LOG_DEBUG, "%8d %d (%d bytes)\n", i, ape_ctx->seektable[i], ape_ctx->seektable[i + 1] - ape_ctx->seektable[i]); + } else { + av_log(NULL, AV_LOG_DEBUG, "%8d %d\n", i, ape_ctx->seektable[i]); + } + } + } + + av_log(NULL, AV_LOG_DEBUG, "\nFrames\n\n"); + for (i = 0; i < ape_ctx->totalframes; i++) + av_log(NULL, AV_LOG_DEBUG, "%8d %8lld %8d (%d samples)\n", i, ape_ctx->frames[i].pos, ape_ctx->frames[i].size, ape_ctx->frames[i].nblocks); + + av_log(NULL, AV_LOG_DEBUG, "\nCalculated information:\n\n"); + av_log(NULL, AV_LOG_DEBUG, "junklength = %d\n", ape_ctx->junklength); + av_log(NULL, AV_LOG_DEBUG, "firstframe = %d\n", ape_ctx->firstframe); + av_log(NULL, AV_LOG_DEBUG, "totalsamples = %d\n", ape_ctx->totalsamples); +} + +static int ape_read_header(AVFormatContext * s, AVFormatParameters * ap) +{ + ByteIOContext *pb = &s->pb; + APEContext *ape = s->priv_data; + AVStream *st; + uint32_t tag; + int i; + int total_blocks; + int64_t pts; + + /* TODO: Skip any leading junk such as id3v2 tags */ + ape->junklength = 0; + + tag = get_le32(pb); + if (tag != MKTAG('M', 'A', 'C', ' ')) + return -1; + + ape->fileversion = get_le16(pb); + + if (ape->fileversion < APE_MIN_VERSION || ape->fileversion > APE_MAX_VERSION) { + av_log(s, AV_LOG_ERROR, "Unsupported file version - %d.%02d\n", ape->fileversion / 1000, (ape->fileversion % 1000) / 10); + return -1; + } + + if (ape->fileversion >= 3980) { + ape->padding1 = get_le16(pb); + ape->descriptorlength = get_le32(pb); + ape->headerlength = get_le32(pb); + ape->seektablelength = get_le32(pb); + ape->wavheaderlength = get_le32(pb); + ape->audiodatalength = get_le32(pb); + ape->audiodatalength_high = get_le32(pb); + ape->wavtaillength = get_le32(pb); + get_buffer(pb, ape->md5, 16); + + /* Skip any unknown bytes at the end of the descriptor. + This is for future compatibility */ + if (ape->descriptorlength > 52) + url_fseek(pb, ape->descriptorlength - 52, SEEK_CUR); + + /* Read header data */ + ape->compressiontype = get_le16(pb); + ape->formatflags = get_le16(pb); + ape->blocksperframe = get_le32(pb); + ape->finalframeblocks = get_le32(pb); + ape->totalframes = get_le32(pb); + ape->bps = get_le16(pb); + ape->channels = get_le16(pb); + ape->samplerate = get_le32(pb); + } else { + ape->descriptorlength = 0; + ape->headerlength = 32; + + ape->compressiontype = get_le16(pb); + ape->formatflags = get_le16(pb); + ape->channels = get_le16(pb); + ape->samplerate = get_le32(pb); + ape->wavheaderlength = get_le32(pb); + ape->wavtaillength = get_le32(pb); + ape->totalframes = get_le32(pb); + ape->finalframeblocks = get_le32(pb); + + if (ape->formatflags & MAC_FORMAT_FLAG_HAS_PEAK_LEVEL) { + url_fseek(pb, 4, SEEK_CUR); /* Skip the peak level */ + ape->headerlength += 4; + } + + if (ape->formatflags & MAC_FORMAT_FLAG_HAS_SEEK_ELEMENTS) { + ape->seektablelength = get_le32(pb); + ape->headerlength += 4; + ape->seektablelength *= sizeof(int32_t); + } else + ape->seektablelength = ape->totalframes * sizeof(int32_t); + + if (ape->formatflags & MAC_FORMAT_FLAG_8_BIT) + ape->bps = 8; + else if (ape->formatflags & MAC_FORMAT_FLAG_24_BIT) + ape->bps = 24; + else + ape->bps = 16; + + if (ape->fileversion >= 3950) + ape->blocksperframe = 73728 * 4; + else if (ape->fileversion >= 3900 || (ape->fileversion >= 3800 && ape->compressiontype >= 4000)) + ape->blocksperframe = 73728; + else + ape->blocksperframe = 9216; + + /* Skip any stored wav header */ + if (!(ape->formatflags & MAC_FORMAT_FLAG_CREATE_WAV_HEADER)) + url_fskip(pb, ape->wavheaderlength); + } + + if(ape->totalframes > UINT_MAX / sizeof(APEFrame)){ + av_log(s, AV_LOG_ERROR, "Too many frames: %d\n", ape->totalframes); + return -1; + } + ape->frames = av_malloc(ape->totalframes * sizeof(APEFrame)); + if(!ape->frames) + return AVERROR_NOMEM; + ape->firstframe = ape->junklength + ape->descriptorlength + ape->headerlength + ape->seektablelength + ape->wavheaderlength; + ape->currentframe = 0; + + + ape->totalsamples = ape->finalframeblocks; + if (ape->totalframes > 1) + ape->totalsamples += ape->blocksperframe * (ape->totalframes - 1); + + if (ape->seektablelength > 0) { + ape->seektable = av_malloc(ape->seektablelength); + for (i = 0; i < ape->seektablelength / sizeof(uint32_t); i++) + ape->seektable[i] = get_le32(pb); + } + + ape->frames[0].pos = ape->firstframe; + ape->frames[0].nblocks = ape->blocksperframe; + ape->frames[0].skip = 0; + for (i = 1; i < ape->totalframes; i++) { + ape->frames[i].pos = ape->seektable[i]; //ape->frames[i-1].pos + ape->blocksperframe; + ape->frames[i].nblocks = ape->blocksperframe; + ape->frames[i - 1].size = ape->frames[i].pos - ape->frames[i - 1].pos; + ape->frames[i].skip = (ape->frames[i].pos - ape->frames[0].pos) & 3; + } + ape->frames[ape->totalframes - 1].size = ape->finalframeblocks * 4; + ape->frames[ape->totalframes - 1].nblocks = ape->finalframeblocks; + + for (i = 0; i < ape->totalframes; i++) { + if(ape->frames[i].skip){ + ape->frames[i].pos -= ape->frames[i].skip; + ape->frames[i].size += ape->frames[i].skip; + } + ape->frames[i].size = (ape->frames[i].size + 3) & ~3; + } + + + ape_dumpinfo(ape); + + av_log(s, AV_LOG_DEBUG, "Decoding file - v%d.%02d, compression level %d\n", ape->fileversion / 1000, (ape->fileversion % 1000) / 10, ape->compressiontype); + + /* now we are ready: build format streams */ + st = av_new_stream(s, 0); + if (!st) + return -1; + + total_blocks = (ape->totalframes == 0) ? 0 : ((ape->totalframes - 1) * ape->blocksperframe) + ape->finalframeblocks; + + st->codec->codec_type = CODEC_TYPE_AUDIO; + st->codec->codec_id = CODEC_ID_APE; + st->codec->codec_tag = MKTAG('A', 'P', 'E', ' '); + st->codec->channels = ape->channels; + st->codec->sample_rate = ape->samplerate; + st->codec->bits_per_sample = ape->bps; + st->codec->frame_size = MAC_SUBFRAME_SIZE; + + st->nb_frames = ape->totalframes; + s->start_time = 0; + s->duration = (int64_t) total_blocks * AV_TIME_BASE / ape->samplerate; + av_set_pts_info(st, 64, MAC_SUBFRAME_SIZE, ape->samplerate); + + st->codec->extradata = av_malloc(APE_EXTRADATA_SIZE); + st->codec->extradata_size = APE_EXTRADATA_SIZE; + AV_WL16(st->codec->extradata + 0, ape->fileversion); + AV_WL16(st->codec->extradata + 2, ape->compressiontype); + AV_WL16(st->codec->extradata + 4, ape->formatflags); + + pts = 0; + for (i = 0; i < ape->totalframes; i++) { + ape->frames[i].pts = pts; + av_add_index_entry(st, ape->frames[i].pos, ape->frames[i].pts, 0, 0, AVINDEX_KEYFRAME); + pts += ape->blocksperframe / MAC_SUBFRAME_SIZE; + } + + return 0; +} + +static int ape_read_packet(AVFormatContext * s, AVPacket * pkt) +{ + int ret; + int nblocks; + APEContext *ape = s->priv_data; + uint32_t extra_size = 8; + + if (url_feof(&s->pb)) + return AVERROR_IO; + if (ape->currentframe > ape->totalframes) + return AVERROR_IO; + + url_fseek (&s->pb, ape->frames[ape->currentframe].pos, SEEK_SET); + + /* Calculate how many blocks there are in this frame */ + if (ape->currentframe == (ape->totalframes - 1)) + nblocks = ape->finalframeblocks; + else + nblocks = ape->blocksperframe; + + if (av_new_packet(pkt, ape->frames[ape->currentframe].size + extra_size) < 0) + return AVERROR_NOMEM; + + AV_WL32(pkt->data , nblocks); + AV_WL32(pkt->data + 4, ape->frames[ape->currentframe].skip); + ret = get_buffer(&s->pb, pkt->data + extra_size, ape->frames[ape->currentframe].size); + + pkt->pts = ape->frames[ape->currentframe].pts; + pkt->stream_index = 0; + + /* note: we need to modify the packet size here to handle the last + packet */ + pkt->size = ret + extra_size; + + ape->currentframe++; + + return 0; +} + +static int ape_read_close(AVFormatContext * s) +{ + APEContext *ape = s->priv_data; + + av_freep(&ape->frames); + av_freep(&ape->seektable); + return 0; +} + +static int ape_read_seek(AVFormatContext *s, int stream_index, int64_t timestamp, int flags) +{ + AVStream *st = s->streams[stream_index]; + APEContext *ape = s->priv_data; + int index = av_index_search_timestamp(st, timestamp, flags); + + if (index < 0) + return -1; + + ape->currentframe = index; + return 0; +} + +AVInputFormat ape_demuxer = { + "ape", + "Monkey's Audio", + sizeof(APEContext), + ape_probe, + ape_read_header, + ape_read_packet, + ape_read_close, + ape_read_seek, + .extensions = "ape,apl,mac" +}; diff --git a/libavformat/avformat.h b/libavformat/avformat.h index 577c215a44..a58200c312 100644 --- a/libavformat/avformat.h +++ b/libavformat/avformat.h @@ -21,8 +21,8 @@ #ifndef AVFORMAT_H #define AVFORMAT_H -#define LIBAVFORMAT_VERSION_INT ((51<<16)+(12<<8)+3) -#define LIBAVFORMAT_VERSION 51.12.3 +#define LIBAVFORMAT_VERSION_INT ((51<<16)+(13<<8)+3) +#define LIBAVFORMAT_VERSION 51.13.3 #define LIBAVFORMAT_BUILD LIBAVFORMAT_VERSION_INT #define LIBAVFORMAT_IDENT "Lavf" AV_STRINGIFY(LIBAVFORMAT_VERSION) |