From d7a473926504e2acfa6ae3bead0938e1f4e03441 Mon Sep 17 00:00:00 2001 From: Paul B Mahol Date: Tue, 2 Oct 2012 13:43:19 +0000 Subject: TAK demuxer, decoder and parser Signed-off-by: Paul B Mahol --- Changelog | 1 + configure | 1 + doc/general.texi | 2 + libavcodec/Makefile | 2 + libavcodec/allcodecs.c | 2 + libavcodec/avcodec.h | 1 + libavcodec/codec_desc.c | 7 + libavcodec/tak.c | 177 ++++++++ libavcodec/tak.h | 151 +++++++ libavcodec/tak_parser.c | 126 ++++++ libavcodec/takdec.c | 972 ++++++++++++++++++++++++++++++++++++++++++ libavcodec/version.h | 2 +- libavformat/Makefile | 1 + libavformat/allformats.c | 1 + libavformat/takdec.c | 184 ++++++++ libavformat/version.h | 4 +- tests/fate/lossless-audio.mak | 3 + tests/ref/fate/lossless-tak | 1 + 18 files changed, 1635 insertions(+), 3 deletions(-) create mode 100644 libavcodec/tak.c create mode 100644 libavcodec/tak.h create mode 100644 libavcodec/tak_parser.c create mode 100644 libavcodec/takdec.c create mode 100644 libavformat/takdec.c create mode 100644 tests/ref/fate/lossless-tak diff --git a/Changelog b/Changelog index a83d714a40..75b5567045 100644 --- a/Changelog +++ b/Changelog @@ -8,6 +8,7 @@ version next: - Opus encoder using libopus - ffprobe -select_streams option - Pinnacle TARGA CineWave YUV16 decoder +- TAK demuxer, decoder and parser version 1.0: diff --git a/configure b/configure index 11519cd736..85fab2dd9e 100755 --- a/configure +++ b/configure @@ -1824,6 +1824,7 @@ sap_muxer_select="rtp_muxer rtp_protocol" sdp_demuxer_select="rtpdec" smoothstreaming_muxer_select="ismv_muxer" spdif_muxer_select="aac_parser" +tak_demuxer_select="tak_parser" tg2_muxer_select="mov_muxer" tgp_muxer_select="mov_muxer" w64_demuxer_deps="wav_demuxer" diff --git a/doc/general.texi b/doc/general.texi index 198c9eef1a..9cba7bb3bb 100644 --- a/doc/general.texi +++ b/doc/general.texi @@ -304,6 +304,7 @@ library: @item raw video @tab X @tab X @item raw id RoQ @tab X @tab @item raw Shorten @tab @tab X +@item raw TAK @tab @tab X @item raw TrueHD @tab X @tab X @item raw VC-1 @tab @tab X @item raw PCM A-law @tab X @tab X @@ -863,6 +864,7 @@ following image formats are supported: @tab experimental codec @item Speex @tab E @tab E @tab supported through external library libspeex +@item TAK (Tom's lossless Audio Kompressor) @tab @tab X @item True Audio (TTA) @tab @tab X @item TrueHD @tab @tab X @tab Used in HD-DVD and Blu-Ray discs. diff --git a/libavcodec/Makefile b/libavcodec/Makefile index 61694fe8d9..7730c16453 100644 --- a/libavcodec/Makefile +++ b/libavcodec/Makefile @@ -400,6 +400,7 @@ OBJS-$(CONFIG_SVQ3_DECODER) += svq3.o svq13.o h263.o h264.o \ h264_loopfilter.o h264_direct.o \ h264_sei.o h264_ps.o h264_refs.o \ h264_cavlc.o h264_cabac.o cabac.o +OBJS-$(CONFIG_TAK_DECODER) += takdec.o tak.o OBJS-$(CONFIG_TARGA_DECODER) += targa.o OBJS-$(CONFIG_TARGA_ENCODER) += targaenc.o rle.o OBJS-$(CONFIG_TARGA_Y216_DECODER) += targa_y216dec.o @@ -713,6 +714,7 @@ OBJS-$(CONFIG_MPEGVIDEO_PARSER) += mpegvideo_parser.o \ OBJS-$(CONFIG_PNM_PARSER) += pnm_parser.o pnm.o OBJS-$(CONFIG_RV30_PARSER) += rv34_parser.o OBJS-$(CONFIG_RV40_PARSER) += rv34_parser.o +OBJS-$(CONFIG_TAK_PARSER) += tak_parser.o tak.o OBJS-$(CONFIG_VC1_PARSER) += vc1_parser.o vc1.o vc1data.o \ msmpeg4.o msmpeg4data.o mpeg4video.o \ h263.o diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c index 039ba9c6a4..ae88fabc0c 100644 --- a/libavcodec/allcodecs.c +++ b/libavcodec/allcodecs.c @@ -328,6 +328,7 @@ void avcodec_register_all(void) REGISTER_DECODER (SMACKAUD, smackaud); REGISTER_ENCDEC (SONIC, sonic); REGISTER_ENCODER (SONIC_LS, sonic_ls); + REGISTER_DECODER (TAK, tak); REGISTER_DECODER (TRUEHD, truehd); REGISTER_DECODER (TRUESPEECH, truespeech); REGISTER_DECODER (TTA, tta); @@ -486,6 +487,7 @@ void avcodec_register_all(void) REGISTER_PARSER (PNM, pnm); REGISTER_PARSER (RV30, rv30); REGISTER_PARSER (RV40, rv40); + REGISTER_PARSER (TAK, tak); REGISTER_PARSER (VC1, vc1); REGISTER_PARSER (VORBIS, vorbis); REGISTER_PARSER (VP3, vp3); diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h index 719921a779..a8fa83423d 100644 --- a/libavcodec/avcodec.h +++ b/libavcodec/avcodec.h @@ -431,6 +431,7 @@ enum AVCodecID { AV_CODEC_ID_SONIC_LS = MKBETAG('S','O','N','L'), AV_CODEC_ID_PAF_AUDIO = MKBETAG('P','A','F','A'), AV_CODEC_ID_OPUS = MKBETAG('O','P','U','S'), + AV_CODEC_ID_TAK = MKBETAG('t','B','a','K'), /* subtitle codecs */ AV_CODEC_ID_FIRST_SUBTITLE = 0x17000, ///< A dummy ID pointing at the start of subtitle codecs. diff --git a/libavcodec/codec_desc.c b/libavcodec/codec_desc.c index c98b2a5882..d1b0b9e3a1 100644 --- a/libavcodec/codec_desc.c +++ b/libavcodec/codec_desc.c @@ -2252,6 +2252,13 @@ static const AVCodecDescriptor codec_descriptors[] = { .long_name = NULL_IF_CONFIG_SMALL("Opus (Opus Interactive Audio Codec)"), .props = AV_CODEC_PROP_LOSSY, }, + { + .id = AV_CODEC_ID_TAK, + .type = AVMEDIA_TYPE_AUDIO, + .name = "tak", + .long_name = NULL_IF_CONFIG_SMALL("TAK (Tom's lossless Audio Kompressor)"), + .props = AV_CODEC_PROP_LOSSLESS, + }, /* subtitle codecs */ { diff --git a/libavcodec/tak.c b/libavcodec/tak.c new file mode 100644 index 0000000000..7e1a047a2c --- /dev/null +++ b/libavcodec/tak.c @@ -0,0 +1,177 @@ +/* + * TAK common code + * Copyright (c) 2012 Paul B Mahol + * + * 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 "tak.h" +#include "libavutil/crc.h" +#include "libavutil/intreadwrite.h" +#include "libavutil/bswap.h" + +static const int64_t tak_channel_layouts[] = { + 0, + AV_CH_FRONT_LEFT, + AV_CH_FRONT_RIGHT, + AV_CH_FRONT_CENTER, + AV_CH_LOW_FREQUENCY, + AV_CH_BACK_LEFT, + AV_CH_BACK_RIGHT, + AV_CH_FRONT_LEFT_OF_CENTER, + AV_CH_FRONT_RIGHT_OF_CENTER, + AV_CH_BACK_CENTER, + AV_CH_SIDE_LEFT, + AV_CH_SIDE_RIGHT, + AV_CH_TOP_CENTER, + AV_CH_TOP_FRONT_LEFT, + AV_CH_TOP_FRONT_CENTER, + AV_CH_TOP_FRONT_RIGHT, + AV_CH_TOP_BACK_LEFT, + AV_CH_TOP_BACK_CENTER, + AV_CH_TOP_BACK_RIGHT, +}; + +static const uint16_t frame_duration_type_quants[] = { + 3, 4, 6, 8, 4096, 8192, 16384, 512, 1024, 2048, +}; + +static int tak_get_nb_samples(int sample_rate, enum TAKFrameSizeType type) +{ + int nb_samples, max_nb_samples; + + if (type <= TAK_FST_250ms) { + nb_samples = sample_rate * frame_duration_type_quants[type] >> + TAK_FRAME_DURATION_QUANT_SHIFT; + max_nb_samples = 16384; + } else if (type < FF_ARRAY_ELEMS(frame_duration_type_quants)) { + nb_samples = frame_duration_type_quants[type]; + max_nb_samples = sample_rate * frame_duration_type_quants[TAK_FST_250ms] >> + TAK_FRAME_DURATION_QUANT_SHIFT; + } else { + return AVERROR_INVALIDDATA; + } + + if (nb_samples <= 0 || nb_samples > max_nb_samples) + return AVERROR_INVALIDDATA; + + return nb_samples; +} + +static int crc_init = 0; +#if CONFIG_SMALL +#define CRC_TABLE_SIZE 257 +#else +#define CRC_TABLE_SIZE 1024 +#endif +static AVCRC crc_24[CRC_TABLE_SIZE]; + +av_cold void ff_tak_init_crc(void) +{ + if (!crc_init) { + av_crc_init(crc_24, 0, 24, 0x864CFBU, sizeof(crc_24)); + crc_init = 1; + } +} + +int ff_tak_check_crc(const uint8_t *buf, unsigned int buf_size) +{ + uint32_t crc, CRC; + + if (buf_size < 4) + return AVERROR_INVALIDDATA; + buf_size -= 3; + + CRC = av_bswap32(AV_RL24(buf + buf_size)) >> 8; + crc = av_crc(crc_24, 0xCE04B7U, buf, buf_size); + if (CRC != crc) + return AVERROR_INVALIDDATA; + + return 0; +} + +void avpriv_tak_parse_streaminfo(GetBitContext *gb, TAKStreamInfo *s) +{ + uint64_t channel_mask = 0; + int frame_type, i; + + s->codec = get_bits(gb, TAK_ENCODER_CODEC_BITS); + skip_bits(gb, TAK_ENCODER_PROFILE_BITS); + + frame_type = get_bits(gb, TAK_SIZE_FRAME_DURATION_BITS); + s->samples = get_bits_longlong(gb, TAK_SIZE_SAMPLES_NUM_BITS); + + s->data_type = get_bits(gb, TAK_FORMAT_DATA_TYPE_BITS); + s->sample_rate = get_bits(gb, TAK_FORMAT_SAMPLE_RATE_BITS) + TAK_SAMPLE_RATE_MIN; + s->bps = get_bits(gb, TAK_FORMAT_BPS_BITS) + TAK_BPS_MIN; + s->channels = get_bits(gb, TAK_FORMAT_CHANNEL_BITS) + TAK_CHANNELS_MIN; + + if (get_bits1(gb)) { + skip_bits(gb, TAK_FORMAT_VALID_BITS); + if (get_bits1(gb)) { + for (i = 0; i < s->channels; i++) { + int value = get_bits(gb, TAK_FORMAT_CH_LAYOUT_BITS); + + if (value < FF_ARRAY_ELEMS(tak_channel_layouts)) + channel_mask |= tak_channel_layouts[value]; + } + } + } + + s->ch_layout = channel_mask; + s->frame_samples = tak_get_nb_samples(s->sample_rate, frame_type); +} + +#define FRAME_IS_LAST 1 +#define FRAME_HAVE_INFO 2 +#define FRAME_HAVE_METADATA 4 + +int ff_tak_decode_frame_header(AVCodecContext *avctx, GetBitContext *gb, + TAKStreamInfo *ti, int log_level_offset) +{ + int flags; + + if (get_bits(gb, TAK_FRAME_HEADER_SYNC_ID_BITS) != TAK_FRAME_HEADER_SYNC_ID) { + av_log(avctx, AV_LOG_ERROR + log_level_offset, "missing sync id\n"); + return AVERROR_INVALIDDATA; + } + + flags = get_bits(gb, TAK_FRAME_HEADER_FLAGS_BITS); + ti->frame_num = get_bits(gb, TAK_FRAME_HEADER_NO_BITS); + + if (flags & FRAME_IS_LAST) { + ti->last_frame_samples = get_bits(gb, TAK_FRAME_HEADER_SAMPLE_COUNT_BITS) + 1; + skip_bits(gb, 2); + } else { + ti->last_frame_samples = 0; + } + + if (flags & FRAME_HAVE_INFO) { + avpriv_tak_parse_streaminfo(gb, ti); + + if (get_bits(gb, 6)) + skip_bits(gb, 25); + align_get_bits(gb); + } + + if (flags & FRAME_HAVE_METADATA) + return AVERROR_INVALIDDATA; + + skip_bits(gb, 24); + + return 0; +} diff --git a/libavcodec/tak.h b/libavcodec/tak.h new file mode 100644 index 0000000000..cfa074defa --- /dev/null +++ b/libavcodec/tak.h @@ -0,0 +1,151 @@ +/* + * TAK decoder/demuxer common code + * Copyright (c) 2012 Paul B Mahol + * + * 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 + */ + +/** + * @file + * TAK (Tom's lossless Audio Kompressor) decoder/demuxer common functions + */ + +#ifndef AVCODEC_TAK_H +#define AVCODEC_TAK_H + +#define BITSTREAM_READER_LE +#include "get_bits.h" +#include "avcodec.h" + +#define TAK_FORMAT_DATA_TYPE_BITS 3 +#define TAK_FORMAT_SAMPLE_RATE_BITS 18 +#define TAK_FORMAT_BPS_BITS 5 +#define TAK_FORMAT_CHANNEL_BITS 4 +#define TAK_FORMAT_VALID_BITS 5 +#define TAK_FORMAT_CH_LAYOUT_BITS 6 +#define TAK_SIZE_FRAME_DURATION_BITS 4 +#define TAK_SIZE_SAMPLES_NUM_BITS 35 +#define TAK_LAST_FRAME_POS_BITS 40 +#define TAK_LAST_FRAME_SIZE_BITS 24 +#define TAK_ENCODER_CODEC_BITS 6 +#define TAK_ENCODER_PROFILE_BITS 4 +#define TAK_ENCODER_VERSION_BITS 24 +#define TAK_SAMPLE_RATE_MIN 6000 +#define TAK_CHANNELS_MIN 1 +#define TAK_BPS_MIN 8 +#define TAK_FRAME_HEADER_FLAGS_BITS 3 +#define TAK_FRAME_HEADER_SYNC_ID 0xA0FF +#define TAK_FRAME_HEADER_SYNC_ID_BITS 16 +#define TAK_FRAME_HEADER_SAMPLE_COUNT_BITS 14 +#define TAK_FRAME_HEADER_NO_BITS 21 +#define TAK_FRAME_DURATION_QUANT_SHIFT 5 +#define TAK_CRC24_BITS 24 + +#define TAK_MAX_CHANNELS ( 1 << TAK_FORMAT_CHANNEL_BITS ) + +#define TAK_MIN_FRAME_HEADER_BITS ( TAK_FRAME_HEADER_SYNC_ID_BITS + \ + TAK_FRAME_HEADER_FLAGS_BITS + \ + TAK_FRAME_HEADER_NO_BITS + \ + TAK_CRC24_BITS ) + +#define TAK_MIN_FRAME_HEADER_LAST_BITS ( TAK_MIN_FRAME_HEADER_BITS + 2 + \ + TAK_FRAME_HEADER_SAMPLE_COUNT_BITS ) + +#define TAK_ENCODER_BITS ( TAK_ENCODER_CODEC_BITS + \ + TAK_ENCODER_PROFILE_BITS ) + +#define TAK_SIZE_BITS ( TAK_SIZE_SAMPLES_NUM_BITS + \ + TAK_SIZE_FRAME_DURATION_BITS ) + +#define TAK_FORMAT_BITS ( TAK_FORMAT_DATA_TYPE_BITS + \ + TAK_FORMAT_SAMPLE_RATE_BITS + \ + TAK_FORMAT_BPS_BITS + \ + TAK_FORMAT_CHANNEL_BITS + 1 + \ + TAK_FORMAT_VALID_BITS + 1 + \ + TAK_FORMAT_CH_LAYOUT_BITS * \ + TAK_MAX_CHANNELS ) + +#define TAK_STREAMINFO_BITS ( TAK_ENCODER_BITS + \ + TAK_SIZE_BITS + \ + TAK_FORMAT_BITS ) + +#define TAK_MAX_FRAME_HEADER_BITS ( TAK_MIN_FRAME_HEADER_LAST_BITS + \ + TAK_STREAMINFO_BITS + 31 ) + +#define TAK_STREAMINFO_BYTES (( TAK_STREAMINFO_BITS + 7 ) / 8) +#define TAK_MAX_FRAME_HEADER_BYTES (( TAK_MAX_FRAME_HEADER_BITS + 7 ) / 8) +#define TAK_MIN_FRAME_HEADER_BYTES (( TAK_MIN_FRAME_HEADER_BITS + 7 ) / 8) + +enum TAKMetaDataType { + TAK_METADATA_END = 0, + TAK_METADATA_STREAMINFO, + TAK_METADATA_SEEKTABLE, + TAK_METADATA_SIMPLE_WAVE_DATA, + TAK_METADATA_ENCODER, + TAK_METADATA_PADDING, + TAK_METADATA_MD5, + TAK_METADATA_LAST_FRAME, +}; + +enum TAKFrameSizeType { + TAK_FST_94ms = 0, + TAK_FST_125ms, + TAK_FST_188ms, + TAK_FST_250ms, + TAK_FST_4096, + TAK_FST_8192, + TAK_FST_16384, + TAK_FST_512, + TAK_FST_1024, + TAK_FST_2048, +}; + +typedef struct TAKStreamInfo { + int codec; + int data_type; + int sample_rate; + int channels; + int bps; + int frame_num; + int frame_samples; + int last_frame_samples; + uint64_t ch_layout; + int64_t samples; +} TAKStreamInfo; + +void ff_tak_init_crc(void); + +int ff_tak_check_crc(const uint8_t *buf, unsigned int buf_size); + +/** + * Parse the Streaminfo metadata block + * @param[in] gb pointer to GetBitContext + * @param[out] s where parsed information is stored + */ +void avpriv_tak_parse_streaminfo(GetBitContext *gb, TAKStreamInfo *s); + +/** + * Validate and decode a frame header. + * @param avctx AVCodecContext to use as av_log() context + * @param[in] gb GetBitContext from which to read frame header + * @param[out] s frame information + * @param log_level_offset log level offset. can be used to silence error messages. + * @return non-zero on error, 0 if ok + */ +int ff_tak_decode_frame_header(AVCodecContext *avctx, GetBitContext *gb, + TAKStreamInfo *s, int log_level_offset); +#endif /* AVCODEC_TAK_H */ diff --git a/libavcodec/tak_parser.c b/libavcodec/tak_parser.c new file mode 100644 index 0000000000..dcc37eca8f --- /dev/null +++ b/libavcodec/tak_parser.c @@ -0,0 +1,126 @@ +/* + * TAK parser + * Copyright (c) 2012 Michael Niedermayer + * + * 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 + */ + +/** + * @file + * TAK parser + **/ + +#include "parser.h" +#include "tak.h" + +typedef struct TAKParseContext { + ParseContext pc; + TAKStreamInfo ti; + int index; +} TAKParseContext; + +static av_cold int tak_init(AVCodecParserContext *s) +{ + ff_tak_init_crc(); + return 0; +} + +static int tak_parse(AVCodecParserContext *s, AVCodecContext *avctx, + const uint8_t **poutbuf, int *poutbuf_size, + const uint8_t *buf, int buf_size) +{ + TAKParseContext *t = s->priv_data; + ParseContext *pc = &t->pc; + int next = END_NOT_FOUND; + GetBitContext gb; + int consumed = 0; + int needed = buf_size ? TAK_MAX_FRAME_HEADER_BYTES : 8; + + if (s->flags & PARSER_FLAG_COMPLETE_FRAMES) { + TAKStreamInfo ti; + init_get_bits(&gb, buf, buf_size); + if (!ff_tak_decode_frame_header(avctx, &gb, &ti, 127)) + s->duration = t->ti.last_frame_samples ? t->ti.last_frame_samples : + t->ti.frame_samples; + *poutbuf = buf; + *poutbuf_size = buf_size; + return buf_size; + } + + while (buf_size || t->index + needed <= pc->index) { + if (buf_size && t->index + TAK_MAX_FRAME_HEADER_BYTES > pc->index) { + int tmp_buf_size = FFMIN(2 * TAK_MAX_FRAME_HEADER_BYTES, buf_size); + const uint8_t *tmp_buf = buf; + + ff_combine_frame(pc, END_NOT_FOUND, &tmp_buf, &tmp_buf_size); + consumed += tmp_buf_size; + buf += tmp_buf_size; + buf_size -= tmp_buf_size; + } + + for (; t->index + needed <= pc->index; t->index++) { + if (pc->buffer[ t->index ] == 0xFF && + pc->buffer[ t->index + 1 ] == 0xA0) { + TAKStreamInfo ti; + + init_get_bits(&gb, pc->buffer + t->index, + 8 * (pc->index - t->index)); + if (!ff_tak_decode_frame_header(avctx, &gb, + pc->frame_start_found ? &ti : &t->ti, 127) && + !ff_tak_check_crc(pc->buffer + t->index, + get_bits_count(&gb) / 8)) { + if (!pc->frame_start_found) { + pc->frame_start_found = 1; + s->duration = t->ti.last_frame_samples ? + t->ti.last_frame_samples : + t->ti.frame_samples; + } else { + pc->frame_start_found = 0; + next = t->index - pc->index; + t->index = 0; + goto found; + } + } + } + } + } +found: + + if (consumed && !buf_size && next == END_NOT_FOUND || + ff_combine_frame(pc, next, &buf, &buf_size) < 0) { + *poutbuf = NULL; + *poutbuf_size = 0; + return buf_size + consumed; + } + + if (next != END_NOT_FOUND) { + next += consumed; + pc->overread = FFMAX(0, -next); + } + + *poutbuf = buf; + *poutbuf_size = buf_size; + return next; +} + +AVCodecParser ff_tak_parser = { + .codec_ids = { AV_CODEC_ID_TAK }, + .priv_data_size = sizeof(TAKParseContext), + .parser_init = tak_init, + .parser_parse = tak_parse, + .parser_close = ff_parse_close, +}; diff --git a/libavcodec/takdec.c b/libavcodec/takdec.c new file mode 100644 index 0000000000..0877b64554 --- /dev/null +++ b/libavcodec/takdec.c @@ -0,0 +1,972 @@ +/* + * TAK decoder + * Copyright (c) 2012 Paul B Mahol + * + * 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 + */ + +/** + * @file + * TAK (Tom's lossless Audio Kompressor) decoder + * @author Paul B Mahol + */ + +#include "tak.h" +#include "avcodec.h" +#include "unary.h" +#include "dsputil.h" + +#define MAX_SUBFRAMES 8 ///< max number of subframes per channel +#define MAX_PREDICTORS 256 + +typedef struct MCDParam { + int8_t present; ///< is decorrelation parameters available for this channel + int8_t index; ///< index into array of decorrelation types + int8_t chan1; + int8_t chan2; +} MCDParam; + +typedef struct TAKDecContext { + AVCodecContext *avctx; ///< parent AVCodecContext + AVFrame frame; ///< AVFrame for decoded output + DSPContext dsp; + TAKStreamInfo ti; + GetBitContext gb; ///< bitstream reader initialized to start at the current frame + + int nb_samples; ///< number of samples in the current frame + int32_t *decode_buffer; + int decode_buffer_size; + int32_t *decoded[TAK_MAX_CHANNELS]; ///< decoded samples for each channel + + int8_t lpc_mode[TAK_MAX_CHANNELS]; + int8_t sample_shift[TAK_MAX_CHANNELS]; ///< shift applied to every sample in the channel + int32_t xred; + int size; + int ared; + int filter_order; + int16_t predictors[MAX_PREDICTORS]; + int nb_subframes; ///< number of subframes in the current frame + int16_t subframe_len[MAX_SUBFRAMES]; ///< subframe length in samples + int subframe_scale; + + int8_t dmode; ///< channel decorrelation type in the current frame + int8_t dshift; + int16_t dfactor; + int8_t dval1; + int8_t dval2; + + MCDParam mcdparams[TAK_MAX_CHANNELS]; ///< multichannel decorrelation parameters + + int wlength; + int uval; + int rval; + int8_t coding_mode[128]; + DECLARE_ALIGNED(16, int16_t, filter)[MAX_PREDICTORS]; + DECLARE_ALIGNED(16, int16_t, residues)[544]; +} TAKDecContext; + +static const int8_t mc_dmodes[] = { + 1, 3, 4, 6, +}; + +static const uint16_t predictor_sizes[] = { + 4, 8, 12, 16, 24, 32, 48, 64, 80, 96, 128, 160, 192, 224, 256, 0, +}; + +static const struct CParam { + int init; + int escape; + int scale; + int aescape; + int bias; +} xcodes[50] = { + { 0x01, 0x0000001, 0x0000001, 0x0000003, 0x0000008 }, + { 0x02, 0x0000003, 0x0000001, 0x0000007, 0x0000006 }, + { 0x03, 0x0000005, 0x0000002, 0x000000E, 0x000000D }, + { 0x03, 0x0000003, 0x0000003, 0x000000D, 0x0000018 }, + { 0x04, 0x000000B, 0x0000004, 0x000001C, 0x0000019 }, + { 0x04, 0x0000006, 0x0000006, 0x000001A, 0x0000030 }, + { 0x05, 0x0000016, 0x0000008, 0x0000038, 0x0000032 }, + { 0x05, 0x000000C, 0x000000C, 0x0000034, 0x0000060 }, + { 0x06, 0x000002C, 0x0000010, 0x0000070, 0x0000064 }, + { 0x06, 0x0000018, 0x0000018, 0x0000068, 0x00000C0 }, + { 0x07, 0x0000058, 0x0000020, 0x00000E0, 0x00000C8 }, + { 0x07, 0x0000030, 0x0000030, 0x00000D0, 0x0000180 }, + { 0x08, 0x00000B0, 0x0000040, 0x00001C0, 0x0000190 }, + { 0x08, 0x0000060, 0x0000060, 0x00001A0, 0x0000300 }, + { 0x09, 0x0000160, 0x0000080, 0x0000380, 0x0000320 }, + { 0x09, 0x00000C0, 0x00000C0, 0x0000340, 0x0000600 }, + { 0x0A, 0x00002C0, 0x0000100, 0x0000700, 0x0000640 }, + { 0x0A, 0x0000180, 0x0000180, 0x0000680, 0x0000C00 }, + { 0x0B, 0x0000580, 0x0000200, 0x0000E00, 0x0000C80 }, + { 0x0B, 0x0000300, 0x0000300, 0x0000D00, 0x0001800 }, + { 0x0C, 0x0000B00, 0x0000400, 0x0001C00, 0x0001900 }, + { 0x0C, 0x0000600, 0x0000600, 0x0001A00, 0x0003000 }, + { 0x0D, 0x0001600, 0x0000800, 0x0003800, 0x0003200 }, + { 0x0D, 0x0000C00, 0x0000C00, 0x0003400, 0x0006000 }, + { 0x0E, 0x0002C00, 0x0001000, 0x0007000, 0x0006400 }, + { 0x0E, 0x0001800, 0x0001800, 0x0006800, 0x000C000 }, + { 0x0F, 0x0005800, 0x0002000, 0x000E000, 0x000C800 }, + { 0x0F, 0x0003000, 0x0003000, 0x000D000, 0x0018000 }, + { 0x10, 0x000B000, 0x0004000, 0x001C000, 0x0019000 }, + { 0x10, 0x0006000, 0x0006000, 0x001A000, 0x0030000 }, + { 0x11, 0x0016000, 0x0008000, 0x0038000, 0x0032000 }, + { 0x11, 0x000C000, 0x000C000, 0x0034000, 0x0060000 }, + { 0x12, 0x002C000, 0x0010000, 0x0070000, 0x0064000 }, + { 0x12, 0x0018000, 0x0018000, 0x0068000, 0x00C0000 }, + { 0x13, 0x0058000, 0x0020000, 0x00E0000, 0x00C8000 }, + { 0x13, 0x0030000, 0x0030000, 0x00D0000, 0x0180000 }, + { 0x14, 0x00B0000, 0x0040000, 0x01C0000, 0x0190000 }, + { 0x14, 0x0060000, 0x0060000, 0x01A0000, 0x0300000 }, + { 0x15, 0x0160000, 0x0080000, 0x0380000, 0x0320000 }, + { 0x15, 0x00C0000, 0x00C0000, 0x0340000, 0x0600000 }, + { 0x16, 0x02C0000, 0x0100000, 0x0700000, 0x0640000 }, + { 0x16, 0x0180000, 0x0180000, 0x0680000, 0x0C00000 }, + { 0x17, 0x0580000, 0x0200000, 0x0E00000, 0x0C80000 }, + { 0x17, 0x0300000, 0x0300000, 0x0D00000, 0x1800000 }, + { 0x18, 0x0B00000, 0x0400000, 0x1C00000, 0x1900000 }, + { 0x18, 0x0600000, 0x0600000, 0x1A00000, 0x3000000 }, + { 0x19, 0x1600000, 0x0800000, 0x3800000, 0x3200000 }, + { 0x19, 0x0C00000, 0x0C00000, 0x3400000, 0x6000000 }, + { 0x1A, 0x2C00000, 0x1000000, 0x7000000, 0x6400000 }, + { 0x1A, 0x1800000, 0x1800000, 0x6800000, 0xC000000 }, +}; + +static void tak_set_bps(AVCodecContext *avctx) +{ + switch (avctx->bits_per_coded_sample) { + 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; + } +} + +static int get_shift(int sample_rate) +{ + int shift; + + if (sample_rate < 11025) + shift = 3; + else if (sample_rate < 22050) + shift = 2; + else if (sample_rate < 44100) + shift = 1; + else + shift = 0; + + return shift; +} + +static int get_scale(int sample_rate, int shift) +{ + return FFALIGN(sample_rate + 511 >> 9, 4) << shift; +} + +static av_cold int tak_decode_init(AVCodecContext *avctx) +{ + TAKDecContext *s = avctx->priv_data; + + ff_tak_init_crc(); + ff_dsputil_init(&s->dsp, avctx); + + s->avctx = avctx; + avcodec_get_frame_defaults(&s->frame); + avctx->coded_frame = &s->frame; + + s->uval = get_scale(avctx->sample_rate, get_shift(avctx->sample_rate)); + s->subframe_scale = get_scale(avctx->sample_rate, 1); + + tak_set_bps(avctx); + + return 0; +} + +static int get_code(GetBitContext *gb, int nbits) +{ + if (nbits == 1) { + skip_bits1(gb); + return 0; + } else { + return get_sbits(gb, nbits); + } +} + +static void decode_lpc(int32_t *coeffs, int mode, int length) +{ + int i, a1, a2, a3, a4, a5; + + if (length < 2) + return; + + if (mode == 1) { + a1 = *coeffs++; + for (i = 0; i < (length - 1 >> 1); i++) { + *coeffs += a1; + coeffs[1] += *coeffs; + a1 = coeffs[1]; + coeffs += 2; + } + if ((length - 1) & 1) + *coeffs += a1; + } else if (mode == 2) { + a1 = coeffs[1]; + a2 = a1 + *coeffs; + coeffs[1] = a2; + if (length > 2) { + coeffs += 2; + for (i = 0; i < (length - 2 >> 1); i++) { + a3 = *coeffs + a1; + a4 = a3 + a2; + *coeffs = a4; + a1 = coeffs[1] + a3; + a2 = a1 + a4; + coeffs[1] = a2; + coeffs += 2; + } + if (length & 1) + *coeffs += a1 + a2; + } + } else if (mode == 3) { + a1 = coeffs[1]; + a2 = a1 + *coeffs; + coeffs[1] = a2; + if (length > 2) { + a3 = coeffs[2]; + a4 = a3 + a1; + a5 = a4 + a2; + coeffs += 3; + for (i = 0; i < length - 3; i++) { + a3 += *coeffs; + a4 += a3; + a5 += a4; + *coeffs = a5; + coeffs++; + } + } + } +} + +static int decode_segment(TAKDecContext *s, int8_t value, int32_t *dst, int len) +{ + GetBitContext *gb = &s->gb; + + if (!value) { + memset(dst, 0, len * 4); + } else { + int x, y, z, i = 0; + + value--; + do { + while (1) { + x = get_bits_long(gb, xcodes[value].init); + if (x >= xcodes[value].escape) + break; + dst[i++] = (x >> 1) ^ -(x & 1); + if (i >= len) + return 0; + } + + y = get_bits1(gb); + x = (y << xcodes[value].init) | x; + if (x >= xcodes[value].aescape) { + int c = get_unary(gb, 1, 9); + + if (c == 9) { + int d; + + z = x + xcodes[value].bias; + d = get_bits(gb, 3); + if (d == 7) { + d = get_bits(gb, 5) + 7; + if (d > 29) + return AVERROR_INVALIDDATA; + } + if (d) + z += xcodes[value].scale * (get_bits_long(gb, d) + 1); + } else { + z = xcodes[value].scale * c + x - xcodes[value].escape; + } + } else { + z = x - (xcodes[value].escape & -y); + } + dst[i++] = (z >> 1) ^ -(z & 1); + } while (i < len); + } + + return 0; +} + +static int xget(TAKDecContext *s, int d, int q) +{ + int x; + + x = d / q; + + s->rval = d - (x * q); + + if (s->rval < q / 2) { + s->rval += q; + } else { + x++; + } + + if (x <= 1 || x > 128) + return -1; + + return x; +} + +static int get_len(TAKDecContext *s, int b) +{ + if (b >= s->wlength - 1) + return s->rval; + else + return s->uval; +} + +static int decode_coeffs(TAKDecContext *s, int32_t *dst, int length) +{ + GetBitContext *gb = &s->gb; + int i, v, ret; + + if (length > s->nb_samples) + return AVERROR_INVALIDDATA; + + if (get_bits1(gb)) { + if ((s->wlength = xget(s, length, s->uval)) < 0) + return AVERROR_INVALIDDATA; + + s->coding_mode[0] = v = get_bits(gb, 6); + if (s->coding_mode[0] > FF_ARRAY_ELEMS(xcodes)) + return AVERROR_INVALIDDATA; + + for (i = 1; i < s->wlength; i++) { + int c = get_unary(gb, 1, 6); + + if (c > 5) { + v = get_bits(gb, 6); + } else if (c > 2) { + int t = get_bits1(gb); + + v += (-t ^ (c - 1)) + t; + } else { + v += (-(c & 1) ^ (((c & 1) + c) >> 1)) + (c & 1); + } + + if (v > FF_ARRAY_ELEMS(xcodes)) + return AVERROR_INVALIDDATA; + s->coding_mode[i] = v; + } + + i = 0; + while (i < s->wlength) { + int len = 0; + + v = s->coding_mode[i]; + do { + len += get_len(s, i); + i++; + + if (i == s->wlength) + break; + } while (v == s->coding_mode[i]); + + if ((ret = decode_segment(s, v, dst, len)) < 0) + return ret; + dst += len; + } + } else { + v = get_bits(gb, 6); + if (v > FF_ARRAY_ELEMS(xcodes)) + return AVERROR_INVALIDDATA; + if ((ret = decode_segment(s, v, dst, length)) < 0) + return ret; + } + + return 0; +} + +static int get_b(GetBitContext *gb) +{ + if (get_bits1(gb)) + return get_bits(gb, 4) + 1; + else + return 0; +} + +static int decode_subframe(TAKDecContext *s, int32_t *ptr, int subframe_size, + int prev_subframe_size) +{ + GetBitContext *gb = &s->gb; + int tmp, x, y, i, j, ret = 0; + int tfilter[MAX_PREDICTORS]; + + if (get_bits1(gb)) { + s->filter_order = predictor_sizes[get_bits(gb, 4)]; + + if (prev_subframe_size > 0 && get_bits1(gb)) { + if (s->filter_order > prev_subframe_size) + return AVERROR_INVALIDDATA; + + ptr -= s->filter_order; + subframe_size += s->filter_order; + + if (s->filter_order > subframe_size) + return AVERROR_INVALIDDATA; + } else { + int lpc; + + if (s->filter_order > subframe_size) + return AVERROR_INVALIDDATA; + + lpc = get_bits(gb, 2); + if (lpc > 2) + return AVERROR_INVALIDDATA; + + if ((ret = decode_coeffs(s, ptr, s->filter_order)) < 0) + return ret; + + decode_lpc(ptr, lpc, s->filter_order); + } + + s->xred = get_b(gb); + s->size = get_bits1(gb) + 5; + + if (get_bits1(gb)) { + s->ared = get_bits(gb, 3) + 1; + if (s->ared > 7) + return AVERROR_INVALIDDATA; + } else { + s->ared = 0; + } + s->predictors[0] = get_code(gb, 10); + s->predictors[1] = get_code(gb, 10); + s->predictors[2] = get_code(gb, s->size + 1) << (9 - s->size); + s->predictors[3] = get_code(gb, s->size + 1) << (9 - s->size); + if (s->filter_order > 4) { + tmp = s->size + 1 - get_bits1(gb); + + for (i = 4; i < s->filter_order; i++) { + if (!(i & 3)) + x = tmp - get_bits(gb, 2); + s->predictors[i] = get_code(gb, x) << (9 - s->size); + } + } + + tfilter[0] = s->predictors[0] << 6; + for (i = 1; i < s->filter_order; i++) { + int32_t *p1 = &tfilter[0]; + int32_t *p2 = &tfilter[i - 1]; + + for (j = 0; j < (i + 1) / 2; j++) { + x = *p1 + (s->predictors[i] * *p2 + 256 >> 9); + *p2 += s->predictors[i] * *p1 + 256 >> 9; + *p1++ = x; + p2--; + } + + tfilter[i] = s->predictors[i] << 6; + } + + x = -1 << (32 - (s->ared + 5)); + y = 1 << ((s->ared + 5) - 1); + for (i = 0, j = s->filter_order - 1; i < s->filter_order / 2; i++, j--) { + tmp = y + tfilter[j]; + s->filter[j] = -(x & -(y + tfilter[i] >> 31) | + (y + tfilter[i]) >> (s->ared + 5)); + s->filter[i] = -(x & -(tmp >> 31) | (tmp >> s->ared + 5)); + } + + if ((ret = decode_coeffs(s, &ptr[s->filter_order], + subframe_size - s->filter_order)) < 0) + return ret; + + for (i = 0; i < s->filter_order; i++) + s->residues[i] = *ptr++ >> s->xred; + + y = FF_ARRAY_ELEMS(s->residues) - s->filter_order; + x = subframe_size - s->filter_order; + while (x > 0) { + tmp = FFMIN(y, x); + + for (i = 0; i < tmp; i++) { + int v, w, m; + + v = 1 << (10 - s->ared - 1); + if (!(s->filter_order & 15)) { + v += s->dsp.scalarproduct_int16(&s->residues[i], s->filter, + s->filter_order); + } else if (s->filter_order & 4) { + for (j = 0; j < s->filter_order; j += 4) { + v += s->residues[i + j + 3] * s->filter[j + 3] + + s->residues[i + j + 2] * s->filter[j + 2] + + s->residues[i + j + 1] * s->filter[j + 1] + + s->residues[i + j ] * s->filter[j ]; + } + } else { + for (j = 0; j < s->filter_order; j += 8) { + v += s->residues[i + j + 7] * s->filter[j + 7] + + s->residues[i + j + 6] * s->filter[j + 6] + + s->residues[i + j + 5] * s->filter[j + 5] + + s->residues[i + j + 4] * s->filter[j + 4] + + s->residues[i + j + 3] * s->filter[j + 3] + + s->residues[i + j + 2] * s->filter[j + 2] + + s->residues[i + j + 1] * s->filter[j + 1] + + s->residues[i + j ] * s->filter[j ]; + } + } + m = (-1 << (32 - (10 - s->ared))) & -(v >> 31) | (v >> 10 - s->ared); + m = av_clip(m, -8192, 8191); + w = (m << s->xred) - *ptr; + *ptr++ = w; + s->residues[s->filter_order + i] = w >> s->xred; + } + + x -= tmp; + if (x > 0) + memcpy(s->residues, &s->residues[y], 2 * s->filter_order); + } + + emms_c(); + } else { + ret = decode_coeffs(s, ptr, subframe_size); + } + + return ret; +} + +static int decode_channel(TAKDecContext *s, int chan) +{ + AVCodecContext *avctx = s->avctx; + GetBitContext *gb = &s->gb; + int32_t *dst = s->decoded[chan]; + int i = 0, ret, prev = 0; + int left = s->nb_samples - 1; + + s->sample_shift[chan] = get_b(gb); + if (s->sample_shift[chan] >= avctx->bits_per_coded_sample) + return AVERROR_INVALIDDATA; + + *dst++ = get_code(gb, avctx->bits_per_coded_sample - s->sample_shift[chan]); + s->lpc_mode[chan] = get_bits(gb, 2); + s->nb_subframes = get_bits(gb, 3) + 1; + + if (s->nb_subframes > 1) { + if (get_bits_left(gb) < (s->nb_subframes - 1) * 6) + return AVERROR_INVALIDDATA; + + for (; i < s->nb_subframes - 1; i++) { + int v = get_bits(gb, 6); + + s->subframe_len[i] = (v - prev) * s->subframe_scale; + if (s->subframe_len[i] <= 0) + return AVERROR_INVALIDDATA; + + left -= s->subframe_len[i]; + prev = v; + } + + if (left <= 0) + return AVERROR_INVALIDDATA; + } + + s->subframe_len[i] = left; + prev = 0; + for (i = 0; i < s->nb_subframes; i++) { + if ((ret = decode_subframe(s, dst, s->subframe_len[i], prev)) < 0) + return ret; + dst += s->subframe_len[i]; + prev = s->subframe_len[i]; + } + + return 0; +} + +static int decorrelate(TAKDecContext *s, int c1, int c2, int length) +{ + GetBitContext *gb = &s->gb; + uint32_t *p1 = s->decoded[c1] + 1; + uint32_t *p2 = s->decoded[c2] + 1; + int a, b, i, x, tmp; + + if (s->dmode > 3) { + s->dshift = get_b(gb); + if (s->dmode > 5) { + if (get_bits1(gb)) + s->filter_order = 16; + else + s->filter_order = 8; + + s->dval1 = get_bits1(gb); + s->dval2 = get_bits1(gb); + + for (i = 0; i < s->filter_order; i++) { + if (!(i & 3)) + x = 14 - get_bits(gb, 3); + s->filter[i] = get_code(gb, x); + } + } else { + s->dfactor = get_code(gb, 10); + } + } + + switch (s->dmode) { + case 1: + for (i = 0; i < length; i++, p1++, p2++) + *p2 += *p1; + break; + case 2: + for (i = 0; i < length; i++, p1++, p2++) + *p1 = *p2 - *p1; + break; + case 3: + for (i = 0; i < length; i++, p1++, p2++) { + x = (*p2 & 1) + 2 * *p1; + a = -*p2 + x; + b = *p2 + x; + *p1 = a & 0x80000000 | (a >> 1); + *p2 = b & 0x80000000 | (b >> 1); + } + break; + case 4: + FFSWAP(uint32_t *, p1, p2); + case 5: + if (s->dshift) + tmp = -1 << (32 - s->dshift); + else + tmp = 0; + + for (i = 0; i < length; i++, p1++, p2++) { + x = s->dfactor * (tmp & -(*p2 >> 31) | (*p2 >> s->dshift)) + 128; + *p1 = ((-(x >> 31) & 0xFF000000 | (x >> 8)) << s->dshift) - *p1; + } + break; + case 6: + FFSWAP(uint32_t *, p1, p2); + case 7: + if (length < 256) + return AVERROR_INVALIDDATA; + + a = s->filter_order / 2; + b = length - (s->filter_order - 1); + + if (s->dval1) { + for (i = 0; i < a; i++) + p1[i] += p2[i]; + } + + if (s->dval2) { + x = a + b; + for (i = 0; i < length - x; i++) + p1[x + i] += p2[x + i]; + } + + for (i = 0; i < s->filter_order; i++) + s->residues[i] = *p2++ >> s->dshift; + + p1 += a; + x = FF_ARRAY_ELEMS(s->residues) - s->filter_order; + for (; b > 0; b -= tmp) { + tmp = FFMIN(b, x); + + for (i = 0; i < tmp; i++) + s->residues[s->filter_order + i] = *p2++ >> s->dshift; + + for (i = 0; i < tmp; i++) { + int v, w, m; + + v = 1 << 9; + + if (s->filter_order == 16) { + v += s->dsp.scalarproduct_int16(&s->residues[i], s->filter, + s->filter_order); + } else { + v += s->residues[i + 7] * s->filter[7] + + s->residues[i + 6] * s->filter[6] + + s->residues[i + 5] * s->filter[5] + + s->residues[i + 4] * s->filter[4] + + s->residues[i + 3] * s->filter[3] + + s->residues[i + 2] * s->filter[2] + + s->residues[i + 1] * s->filter[1] + + s->residues[i ] * s->filter[0]; + } + + m = (-1 << 22) & -(v >> 31) | (v >> 10); + m = av_clip(m, -8192, 8191); + w = (m << s->dshift) - *p1; + *p1++ = w; + } + + memcpy(s->residues, &s->residues[tmp], 2 * s->filter_order); + } + + emms_c(); + break; + } + + return 0; +} + +static int tak_decode_frame(AVCodecContext *avctx, void *data, + int *got_frame_ptr, AVPacket *pkt) +{ + TAKDecContext *s = avctx->priv_data; + GetBitContext *gb = &s->gb; + int chan, i, ret, hsize; + int32_t *p; + + if (pkt->size < TAK_MIN_FRAME_HEADER_BYTES) + return AVERROR_INVALIDDATA; + + init_get_bits(gb, pkt->data, pkt->size * 8); + + if ((ret = ff_tak_decode_frame_header(avctx, gb, &s->ti, 0)) < 0) + return ret; + + if (avctx->err_recognition & AV_EF_CRCCHECK) { + hsize = get_bits_count(gb) / 8; + if (ff_tak_check_crc(pkt->data, hsize)) { + av_log(avctx, AV_LOG_ERROR, "CRC error\n"); + return AVERROR_INVALIDDATA; + } + } + + if (s->ti.codec != 2 && s->ti.codec != 4) { + av_log(avctx, AV_LOG_ERROR, "unsupported codec: %d\n", s->ti.codec); + return AVERROR_PATCHWELCOME; + } + if (s->ti.data_type) { + av_log(avctx, AV_LOG_ERROR, "unsupported data type: %d\n", s->ti.data_type); + return AVERROR_INVALIDDATA; + } + if (s->ti.codec == 2 && s->ti.channels > 2) { + av_log(avctx, AV_LOG_ERROR, "invalid number of channels: %d\n", s->ti.channels); + return AVERROR_INVALIDDATA; + } + if (s->ti.channels > 6) { + av_log(avctx, AV_LOG_ERROR, "unsupported number of channels: %d\n", s->ti.channels); + return AVERROR_INVALIDDATA; + } + + if (s->ti.frame_samples <= 0) { + av_log(avctx, AV_LOG_ERROR, "unsupported/invalid number of samples\n"); + return AVERROR_INVALIDDATA; + } + + if (s->ti.bps != avctx->bits_per_coded_sample) { + avctx->bits_per_coded_sample = s->ti.bps; + tak_set_bps(avctx); + } + if (s->ti.sample_rate != avctx->sample_rate) { + avctx->sample_rate = s->ti.sample_rate; + s->uval = get_scale(avctx->sample_rate, get_shift(avctx->sample_rate)); + s->subframe_scale = get_scale(avctx->sample_rate, 1); + } + if (s->ti.ch_layout) + avctx->channel_layout = s->ti.ch_layout; + avctx->channels = s->ti.channels; + + s->nb_samples = s->ti.last_frame_samples ? s->ti.last_frame_samples : + s->ti.frame_samples; + + s->frame.nb_samples = s->nb_samples; + if ((ret = avctx->get_buffer(avctx, &s->frame)) < 0) + return ret; + + if (avctx->bits_per_coded_sample <= 16) { + av_fast_malloc(&s->decode_buffer, &s->decode_buffer_size, + sizeof(*s->decode_buffer) * FFALIGN(s->nb_samples, 8) * + avctx->channels + FF_INPUT_BUFFER_PADDING_SIZE); + if (!s->decode_buffer) + return AVERROR(ENOMEM); + for (chan = 0; chan < avctx->channels; chan++) + s->decoded[chan] = s->decode_buffer + + chan * FFALIGN(s->nb_samples, 8); + } else { + for (chan = 0; chan < avctx->channels; chan++) + s->decoded[chan] = (int32_t *)s->frame.data[chan]; + } + + if (s->nb_samples < 16) { + for (chan = 0; chan < avctx->channels; chan++) { + p = s->decoded[chan]; + for (i = 0; i < s->nb_samples; i++) + *p++ = get_code(gb, avctx->bits_per_coded_sample); + } + } else { + if (s->ti.codec == 2) { + for (chan = 0; chan < avctx->channels; chan++) { + if (ret = decode_channel(s, chan)) + return ret; + } + + if (avctx->channels == 2) { + s->nb_subframes = get_bits(gb, 1) + 1; + if (s->nb_subframes > 1) + s->subframe_len[1] = get_bits(gb, 6); + + s->dmode = get_bits(gb, 3); + if (ret = decorrelate(s, 0, 1, s->nb_samples - 1)) + return ret; + } + } else if (s->ti.codec == 4) { + if (get_bits1(gb)) { + int ch_mask = 0; + + chan = get_bits(gb, 4) + 1; + if (chan > avctx->channels) + return AVERROR_INVALIDDATA; + + for (i = 0; i < chan; i++) { + int nbit = get_bits(gb, 4); + + if (nbit >= avctx->channels) + return AVERROR_INVALIDDATA; + + if (ch_mask & 1 << nbit) + return AVERROR_INVALIDDATA; + + s->mcdparams[i].present = get_bits1(gb); + if (s->mcdparams[i].present) { + s->mcdparams[i].index = get_bits(gb, 2); + s->mcdparams[i].chan2 = get_bits(gb, 4); + if (s->mcdparams[i].index == 1) { + if ((nbit == s->mcdparams[i].chan2) || + (ch_mask & 1 << s->mcdparams[i].chan2)) + return AVERROR_INVALIDDATA; + + ch_mask |= 1 << s->mcdparams[i].chan2; + } else if (!(ch_mask & 1 << s->mcdparams[i].chan2)) { + return AVERROR_INVALIDDATA; + } + } + s->mcdparams[i].chan1 = nbit; + + ch_mask |= 1 << nbit; + } + } else { + chan = avctx->channels; + for (i = 0; i < chan; i++) { + s->mcdparams[i].present = 0; + s->mcdparams[i].chan1 = i; + } + } + + for (i = 0; i < chan; i++) { + if (s->mcdparams[i].present && s->mcdparams[i].index == 1) { + if (ret = decode_channel(s, s->mcdparams[i].chan2)) + return ret; + } + + if (ret = decode_channel(s, s->mcdparams[i].chan1)) + return ret; + + if (s->mcdparams[i].present) { + s->dmode = mc_dmodes[s->mcdparams[i].index]; + if (ret = decorrelate(s, s->mcdparams[i].chan2, + s->mcdparams[i].chan1, + s->nb_samples - 1)) + return ret; + } + } + } + + for (chan = 0; chan < avctx->channels; chan++) { + p = s->decoded[chan]; + decode_lpc(p, s->lpc_mode[chan], s->nb_samples); + + if (s->sample_shift[chan] > 0) { + for (i = 0; i < s->nb_samples; i++) + *p++ <<= s->sample_shift[chan]; + } + } + } + + align_get_bits(gb); + skip_bits(gb, 24); + if (get_bits_left(gb) < 0) + av_log(avctx, AV_LOG_DEBUG, "overread\n"); + else if (get_bits_left(gb) > 0) + av_log(avctx, AV_LOG_DEBUG, "underread\n"); + + if (avctx->err_recognition & AV_EF_CRCCHECK) { + if (ff_tak_check_crc(pkt->data + hsize, + get_bits_count(gb) / 8 - hsize)) { + av_log(avctx, AV_LOG_ERROR, "CRC error\n"); + return AVERROR_INVALIDDATA; + } + } + + // convert to output buffer + switch (avctx->bits_per_coded_sample) { + case 8: + for (chan = 0; chan < avctx->channels; chan++) { + uint8_t *samples = (uint8_t *)s->frame.data[chan]; + p = s->decoded[chan]; + for (i = 0; i < s->nb_samples; i++, p++) + *samples++ = *p + 0x80; + } + break; + case 16: + for (chan = 0; chan < avctx->channels; chan++) { + int16_t *samples = (int16_t *)s->frame.data[chan]; + p = s->decoded[chan]; + for (i = 0; i < s->nb_samples; i++, p++) + *samples++ = *p; + } + break; + case 24: + for (chan = 0; chan < avctx->channels; chan++) { + int32_t *samples = (int32_t *)s->frame.data[chan]; + for (i = 0; i < s->nb_samples; i++) + *samples++ <<= 8; + } + break; + } + + *got_frame_ptr = 1; + *(AVFrame *)data = s->frame; + + return pkt->size; +} + +static av_cold int tak_decode_close(AVCodecContext *avctx) +{ + TAKDecContext *s = avctx->priv_data; + + av_freep(&s->decode_buffer); + + return 0; +} + +AVCodec ff_tak_decoder = { + .name = "tak", + .type = AVMEDIA_TYPE_AUDIO, + .id = AV_CODEC_ID_TAK, + .priv_data_size = sizeof(TAKDecContext), + .init = tak_decode_init, + .close = tak_decode_close, + .decode = tak_decode_frame, + .capabilities = CODEC_CAP_DR1, + .long_name = NULL_IF_CONFIG_SMALL("TAK (Tom's lossless Audio Kompressor)"), +}; diff --git a/libavcodec/version.h b/libavcodec/version.h index b2f4bb010f..506f9f83f6 100644 --- a/libavcodec/version.h +++ b/libavcodec/version.h @@ -29,7 +29,7 @@ #include "libavutil/avutil.h" #define LIBAVCODEC_VERSION_MAJOR 54 -#define LIBAVCODEC_VERSION_MINOR 64 +#define LIBAVCODEC_VERSION_MINOR 65 #define LIBAVCODEC_VERSION_MICRO 100 #define LIBAVCODEC_VERSION_INT AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \ diff --git a/libavformat/Makefile b/libavformat/Makefile index 9481d2bd12..1fd616bebb 100644 --- a/libavformat/Makefile +++ b/libavformat/Makefile @@ -333,6 +333,7 @@ OBJS-$(CONFIG_STR_DEMUXER) += psxstr.o OBJS-$(CONFIG_SUBVIEWER_DEMUXER) += subviewerdec.o OBJS-$(CONFIG_SWF_DEMUXER) += swfdec.o swf.o OBJS-$(CONFIG_SWF_MUXER) += swfenc.o swf.o +OBJS-$(CONFIG_TAK_DEMUXER) += takdec.o apetag.o img2.o rawdec.o OBJS-$(CONFIG_THP_DEMUXER) += thp.o OBJS-$(CONFIG_TIERTEXSEQ_DEMUXER) += tiertexseq.o OBJS-$(CONFIG_MKVTIMESTAMP_V2_MUXER) += mkvtimestamp_v2.o diff --git a/libavformat/allformats.c b/libavformat/allformats.c index 36b6988f11..f1e8edcd65 100644 --- a/libavformat/allformats.c +++ b/libavformat/allformats.c @@ -233,6 +233,7 @@ void av_register_all(void) REGISTER_DEMUXER (STR, str); REGISTER_DEMUXER (SUBVIEWER, subviewer); REGISTER_MUXDEMUX (SWF, swf); + REGISTER_DEMUXER (TAK, tak); REGISTER_MUXER (TG2, tg2); REGISTER_MUXER (TGP, tgp); REGISTER_DEMUXER (THP, thp); diff --git a/libavformat/takdec.c b/libavformat/takdec.c new file mode 100644 index 0000000000..6fd3d93cbd --- /dev/null +++ b/libavformat/takdec.c @@ -0,0 +1,184 @@ +/* + * Raw TAK demuxer + * Copyright (c) 2012 Paul B Mahol + * + * 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 "libavcodec/tak.h" +#include "avformat.h" +#include "internal.h" +#include "rawdec.h" +#include "apetag.h" + +typedef struct TAKDemuxContext { + int mlast_frame; + int64_t left; +} TAKDemuxContext; + +static int tak_probe(AVProbeData *p) +{ + if (!memcmp(p->buf, "tBaK", 4)) + return AVPROBE_SCORE_MAX / 2; + return 0; +} + +static int tak_read_header(AVFormatContext *s) +{ + TAKDemuxContext *tc = s->priv_data; + AVIOContext *pb = s->pb; + GetBitContext gb; + AVStream *st; + uint8_t *buffer = NULL; + int ret; + + st = avformat_new_stream(s, 0); + if (!st) + return AVERROR(ENOMEM); + + st->codec->codec_type = AVMEDIA_TYPE_AUDIO; + st->codec->codec_id = AV_CODEC_ID_TAK; + st->need_parsing = AVSTREAM_PARSE_FULL_RAW; + + tc->mlast_frame = 0; + if (avio_rl32(pb) != MKTAG('t', 'B', 'a', 'K')) { + avio_seek(pb, -4, SEEK_CUR); + return 0; + } + + while (!url_feof(pb)) { + enum TAKMetaDataType type; + int size; + + type = avio_r8(pb) & 0x7f; + size = avio_rl24(pb); + + switch (type) { + case TAK_METADATA_STREAMINFO: + case TAK_METADATA_LAST_FRAME: + case TAK_METADATA_ENCODER: + buffer = av_malloc(size + FF_INPUT_BUFFER_PADDING_SIZE); + if (!buffer) + return AVERROR(ENOMEM); + + if (avio_read(pb, buffer, size) != size) { + av_freep(&buffer); + return AVERROR(EIO); + } + + init_get_bits(&gb, buffer, size * 8); + break; + case TAK_METADATA_MD5: { + uint8_t md5[16]; + int i; + + if (size != 19) + return AVERROR_INVALIDDATA; + avio_read(pb, md5, 16); + avio_skip(pb, 3); + av_log(s, AV_LOG_VERBOSE, "MD5="); + for (i = 0; i < 16; i++) + av_log(s, AV_LOG_VERBOSE, "%02x", md5[i]); + av_log(s, AV_LOG_VERBOSE, "\n"); + break; + } + case TAK_METADATA_END: + if (pb->seekable) { + int64_t curpos = avio_tell(pb); + + ff_ape_parse_tag(s); + avio_seek(pb, curpos, SEEK_SET); + } + return 0; + break; + default: + ret = avio_skip(pb, size); + if (ret < 0) + return ret; + } + + if (type == TAK_METADATA_STREAMINFO) { + TAKStreamInfo ti; + + avpriv_tak_parse_streaminfo(&gb, &ti); + if (ti.samples > 0) + st->duration = ti.samples; + st->codec->bits_per_coded_sample = ti.bps; + if (ti.ch_layout) + st->codec->channel_layout = ti.ch_layout; + st->codec->sample_rate = ti.sample_rate; + st->codec->channels = ti.channels; + st->start_time = 0; + avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate); + st->codec->extradata = buffer; + st->codec->extradata_size = size; + buffer = NULL; + } else if (type == TAK_METADATA_LAST_FRAME) { + if (size != 11) + return AVERROR_INVALIDDATA; + tc->mlast_frame = 1; + tc->left = get_bits_longlong(&gb, TAK_LAST_FRAME_POS_BITS) + + get_bits(&gb, TAK_LAST_FRAME_SIZE_BITS); + av_freep(&buffer); + } else if (type == TAK_METADATA_ENCODER) { + av_log(s, AV_LOG_VERBOSE, "encoder version: %0X\n", + get_bits_long(&gb, TAK_ENCODER_VERSION_BITS)); + av_freep(&buffer); + } + } + + return AVERROR_EOF; +} + +static int raw_read_packet(AVFormatContext *s, AVPacket *pkt) +{ + TAKDemuxContext *tc = s->priv_data; + int ret; + + if (tc->mlast_frame) { + AVIOContext *pb = s->pb; + int64_t size; + + size = FFMIN(tc->left, 1024); + if (!size) + return AVERROR_EOF; + + ret = av_get_packet(pb, pkt, size); + if (ret < 0) + return ret; + + pkt->stream_index = 0; + pkt->pos = avio_tell(pb); + tc->left -= ret; + } else { + ret = ff_raw_read_partial_packet(s, pkt); + } + + return ret; +} + +AVInputFormat ff_tak_demuxer = { + .name = "tak", + .long_name = NULL_IF_CONFIG_SMALL("raw TAK"), + .priv_data_size = sizeof(TAKDemuxContext), + .read_probe = tak_probe, + .read_header = tak_read_header, + .read_packet = raw_read_packet, + .flags = AVFMT_GENERIC_INDEX, + .extensions = "tak", + .raw_codec_id = AV_CODEC_ID_TAK, +}; diff --git a/libavformat/version.h b/libavformat/version.h index 32a70d2792..b733ef6c3c 100644 --- a/libavformat/version.h +++ b/libavformat/version.h @@ -30,8 +30,8 @@ #include "libavutil/avutil.h" #define LIBAVFORMAT_VERSION_MAJOR 54 -#define LIBAVFORMAT_VERSION_MINOR 29 -#define LIBAVFORMAT_VERSION_MICRO 105 +#define LIBAVFORMAT_VERSION_MINOR 30 +#define LIBAVFORMAT_VERSION_MICRO 100 #define LIBAVFORMAT_VERSION_INT AV_VERSION_INT(LIBAVFORMAT_VERSION_MAJOR, \ LIBAVFORMAT_VERSION_MINOR, \ diff --git a/tests/fate/lossless-audio.mak b/tests/fate/lossless-audio.mak index e87bb995ea..d511596ac9 100644 --- a/tests/fate/lossless-audio.mak +++ b/tests/fate/lossless-audio.mak @@ -10,6 +10,9 @@ fate-lossless-monkeysaudio: CMD = md5 -i $(SAMPLES)/lossless-audio/luckynight-pa FATE_SAMPLES_LOSSLESS_AUDIO += fate-lossless-shorten fate-lossless-shorten: CMD = md5 -i $(SAMPLES)/lossless-audio/luckynight-partial.shn -f s16le +FATE_SAMPLES_LOSSLESS_AUDIO += fate-lossless-tak +fate-lossless-tak: CMD = crc -i $(SAMPLES)/lossless-audio/luckynight-partial.tak -f s16le + FATE_SAMPLES_LOSSLESS_AUDIO += fate-lossless-tta fate-lossless-tta: CMD = crc -i $(SAMPLES)/lossless-audio/inside.tta diff --git a/tests/ref/fate/lossless-tak b/tests/ref/fate/lossless-tak new file mode 100644 index 0000000000..9e9ba47652 --- /dev/null +++ b/tests/ref/fate/lossless-tak @@ -0,0 +1 @@ +CRC=0x4ec0971f -- cgit v1.2.3