X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fbinkaudio.c;h=eb9cd0ab62c175bae60dcbedc38e178b3dc84af2;hb=f963f80399deb1a2b44c1bac3af7123e8a0c9e46;hp=93adf1ced31b7bbaee9587e2b547956d615f4be2;hpb=2912e87a6c9264d556734e2bf94a99c64cf9b102;p=ffmpeg diff --git a/libavcodec/binkaudio.c b/libavcodec/binkaudio.c index 93adf1ced31..eb9cd0ab62c 100644 --- a/libavcodec/binkaudio.c +++ b/libavcodec/binkaudio.c @@ -28,23 +28,24 @@ * http://wiki.multimedia.cx/index.php?title=Bink_Audio */ +#include "libavutil/channel_layout.h" #include "avcodec.h" -#define ALT_BITSTREAM_READER_LE +#define BITSTREAM_READER_LE #include "get_bits.h" -#include "dsputil.h" -#include "fft.h" +#include "dct.h" +#include "rdft.h" #include "fmtconvert.h" -#include "libavutil/intfloat_readwrite.h" +#include "internal.h" +#include "wma_freqs.h" +#include "libavutil/intfloat.h" -extern const uint16_t ff_wma_critical_freqs[25]; +static float quant_table[96]; #define MAX_CHANNELS 2 #define BINK_BLOCK_MAX_SIZE (MAX_CHANNELS << 11) typedef struct { GetBitContext gb; - DSPContext dsp; - FmtConvertContext fmt_conv; int version_b; ///< Bink version 'b' int first; int channels; @@ -54,9 +55,9 @@ typedef struct { int num_bands; unsigned int *bands; float root; - DECLARE_ALIGNED(16, FFTSample, coeffs)[BINK_BLOCK_MAX_SIZE]; - DECLARE_ALIGNED(16, short, previous)[BINK_BLOCK_MAX_SIZE / 16]; ///< coeffs from previous audio block - float *coeffs_ptr[MAX_CHANNELS]; ///< pointers to the coeffs arrays for float_to_int16_interleave + DECLARE_ALIGNED(32, FFTSample, coeffs)[BINK_BLOCK_MAX_SIZE]; + float previous[MAX_CHANNELS][BINK_BLOCK_MAX_SIZE / 16]; ///< coeffs from previous audio block + uint8_t *packet_buffer; union { RDFTContext rdft; DCTContext dct; @@ -72,9 +73,6 @@ static av_cold int decode_init(AVCodecContext *avctx) int i; int frame_len_bits; - dsputil_init(&s->dsp, avctx); - ff_fmt_convert_init(&s->fmt_conv, avctx); - /* determine frame length */ if (avctx->sample_rate < 22050) { frame_len_bits = 9; @@ -88,24 +86,35 @@ static av_cold int decode_init(AVCodecContext *avctx) av_log(avctx, AV_LOG_ERROR, "too many channels: %d\n", avctx->channels); return -1; } + avctx->channel_layout = avctx->channels == 1 ? AV_CH_LAYOUT_MONO : + AV_CH_LAYOUT_STEREO; - s->version_b = avctx->codec_tag == MKTAG('B','I','K','b'); + s->version_b = avctx->extradata && avctx->extradata[3] == 'b'; - if (avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT) { + if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) { // audio is already interleaved for the RDFT format variant + avctx->sample_fmt = AV_SAMPLE_FMT_FLT; sample_rate *= avctx->channels; s->channels = 1; if (!s->version_b) frame_len_bits += av_log2(avctx->channels); } else { s->channels = avctx->channels; + avctx->sample_fmt = AV_SAMPLE_FMT_FLTP; } s->frame_len = 1 << frame_len_bits; s->overlap_len = s->frame_len / 16; s->block_size = (s->frame_len - s->overlap_len) * s->channels; sample_rate_half = (sample_rate + 1) / 2; - s->root = 2.0 / sqrt(s->frame_len); + if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) + s->root = 2.0 / (sqrt(s->frame_len) * 32768.0); + else + s->root = s->frame_len / (sqrt(s->frame_len) * 32768.0); + for (i = 0; i < 96; i++) { + /* constant is result of 0.066399999/log10(M_E) */ + quant_table[i] = expf(i * 0.15289164787221953823f) * s->root; + } /* calculate number of bands */ for (s->num_bands = 1; s->num_bands < 25; s->num_bands++) @@ -123,12 +132,8 @@ static av_cold int decode_init(AVCodecContext *avctx) s->bands[s->num_bands] = s->frame_len; s->first = 1; - avctx->sample_fmt = AV_SAMPLE_FMT_S16; - for (i = 0; i < s->channels; i++) - s->coeffs_ptr[i] = s->coeffs + i * s->frame_len; - - if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT) + if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) ff_rdft_init(&s->trans.rdft, frame_len_bits, DFT_C2R); else if (CONFIG_BINKAUDIO_DCT_DECODER) ff_dct_init(&s->trans.dct, frame_len_bits, DCT_III); @@ -154,8 +159,9 @@ static const uint8_t rle_length_tab[16] = { /** * Decode Bink Audio block * @param[out] out Output buffer (must contain s->block_size elements) + * @return 0 on success, negative error code on failure */ -static void decode_block(BinkAudioContext *s, short *out, int use_dct) +static int decode_block(BinkAudioContext *s, float **out, int use_dct) { int ch, i, j, k; float q, quant[25]; @@ -166,19 +172,25 @@ static void decode_block(BinkAudioContext *s, short *out, int use_dct) skip_bits(gb, 2); for (ch = 0; ch < s->channels; ch++) { - FFTSample *coeffs = s->coeffs_ptr[ch]; + FFTSample *coeffs = out[ch]; + if (s->version_b) { - coeffs[0] = av_int2flt(get_bits(gb, 32)) * s->root; - coeffs[1] = av_int2flt(get_bits(gb, 32)) * s->root; + if (get_bits_left(gb) < 64) + return AVERROR_INVALIDDATA; + coeffs[0] = av_int2float(get_bits_long(gb, 32)) * s->root; + coeffs[1] = av_int2float(get_bits_long(gb, 32)) * s->root; } else { + if (get_bits_left(gb) < 58) + return AVERROR_INVALIDDATA; coeffs[0] = get_float(gb) * s->root; coeffs[1] = get_float(gb) * s->root; } + if (get_bits_left(gb) < s->num_bands * 8) + return AVERROR_INVALIDDATA; for (i = 0; i < s->num_bands; i++) { - /* constant is result of 0.066399999/log10(M_E) */ int value = get_bits(gb, 8); - quant[i] = expf(FFMIN(value, 95) * 0.15289164787221953823f) * s->root; + quant[i] = quant_table[FFMIN(value, 95)]; } k = 0; @@ -189,10 +201,14 @@ static void decode_block(BinkAudioContext *s, short *out, int use_dct) while (i < s->frame_len) { if (s->version_b) { j = i + 16; - } else if (get_bits1(gb)) { - j = i + rle_length_tab[get_bits(gb, 4)] * 8; } else { - j = i + 8; + int v = get_bits1(gb); + if (v) { + v = get_bits(gb, 4); + j = i + rle_length_tab[v] * 8; + } else { + j = i + 8; + } } j = FFMIN(j, s->frame_len); @@ -209,7 +225,9 @@ static void decode_block(BinkAudioContext *s, short *out, int use_dct) q = quant[k++]; coeff = get_bits(gb, width); if (coeff) { - if (get_bits1(gb)) + int v; + v = get_bits1(gb); + if (v) coeffs[i] = -q * coeff; else coeffs[i] = q * coeff; @@ -223,38 +241,40 @@ static void decode_block(BinkAudioContext *s, short *out, int use_dct) if (CONFIG_BINKAUDIO_DCT_DECODER && use_dct) { coeffs[0] /= 0.5; - ff_dct_calc (&s->trans.dct, coeffs); - s->dsp.vector_fmul_scalar(coeffs, coeffs, s->frame_len / 2, s->frame_len); + s->trans.dct.dct_calc(&s->trans.dct, coeffs); } else if (CONFIG_BINKAUDIO_RDFT_DECODER) - ff_rdft_calc(&s->trans.rdft, coeffs); + s->trans.rdft.rdft_calc(&s->trans.rdft, coeffs); } - s->fmt_conv.float_to_int16_interleave(out, (const float **)s->coeffs_ptr, - s->frame_len, s->channels); - - if (!s->first) { + for (ch = 0; ch < s->channels; ch++) { + int j; int count = s->overlap_len * s->channels; - int shift = av_log2(count); - for (i = 0; i < count; i++) { - out[i] = (s->previous[i] * (count - i) + out[i] * i) >> shift; + if (!s->first) { + j = ch; + for (i = 0; i < s->overlap_len; i++, j += s->channels) + out[ch][i] = (s->previous[ch][i] * (count - j) + + out[ch][i] * j) / count; } + memcpy(s->previous[ch], &out[ch][s->frame_len - s->overlap_len], + s->overlap_len * sizeof(*s->previous[ch])); } - memcpy(s->previous, out + s->block_size, - s->overlap_len * s->channels * sizeof(*out)); - s->first = 0; + + return 0; } static av_cold int decode_end(AVCodecContext *avctx) { BinkAudioContext * s = avctx->priv_data; av_freep(&s->bands); - if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT) + av_freep(&s->packet_buffer); + if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT) ff_rdft_end(&s->trans.rdft); else if (CONFIG_BINKAUDIO_DCT_DECODER) ff_dct_end(&s->trans.dct); + return 0; } @@ -264,52 +284,77 @@ static void get_bits_align32(GetBitContext *s) if (n) skip_bits(s, n); } -static int decode_frame(AVCodecContext *avctx, - void *data, int *data_size, - AVPacket *avpkt) +static int decode_frame(AVCodecContext *avctx, void *data, + int *got_frame_ptr, AVPacket *avpkt) { BinkAudioContext *s = avctx->priv_data; - const uint8_t *buf = avpkt->data; - int buf_size = avpkt->size; - short *samples = data; - short *samples_end = (short*)((uint8_t*)data + *data_size); - int reported_size; + AVFrame *frame = data; GetBitContext *gb = &s->gb; + int ret, consumed = 0; + + if (!get_bits_left(gb)) { + uint8_t *buf; + /* handle end-of-stream */ + if (!avpkt->size) { + *got_frame_ptr = 0; + return 0; + } + if (avpkt->size < 4) { + av_log(avctx, AV_LOG_ERROR, "Packet is too small\n"); + return AVERROR_INVALIDDATA; + } + buf = av_realloc(s->packet_buffer, avpkt->size + FF_INPUT_BUFFER_PADDING_SIZE); + if (!buf) + return AVERROR(ENOMEM); + s->packet_buffer = buf; + memcpy(s->packet_buffer, avpkt->data, avpkt->size); + init_get_bits(gb, s->packet_buffer, avpkt->size * 8); + consumed = avpkt->size; + + /* skip reported size */ + skip_bits_long(gb, 32); + } - init_get_bits(gb, buf, buf_size * 8); + /* get output buffer */ + frame->nb_samples = s->frame_len; + if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) { + av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n"); + return ret; + } - reported_size = get_bits_long(gb, 32); - while (get_bits_count(gb) / 8 < buf_size && - samples + s->block_size <= samples_end) { - decode_block(s, samples, avctx->codec->id == CODEC_ID_BINKAUDIO_DCT); - samples += s->block_size; - get_bits_align32(gb); + if (decode_block(s, (float **)frame->extended_data, + avctx->codec->id == AV_CODEC_ID_BINKAUDIO_DCT)) { + av_log(avctx, AV_LOG_ERROR, "Incomplete packet\n"); + return AVERROR_INVALIDDATA; } + get_bits_align32(gb); + + frame->nb_samples = s->block_size / avctx->channels; + *got_frame_ptr = 1; - *data_size = FFMIN(reported_size, (uint8_t*)samples - (uint8_t*)data); - return buf_size; + return consumed; } AVCodec ff_binkaudio_rdft_decoder = { - "binkaudio_rdft", - AVMEDIA_TYPE_AUDIO, - CODEC_ID_BINKAUDIO_RDFT, - sizeof(BinkAudioContext), - decode_init, - NULL, - decode_end, - decode_frame, - .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (RDFT)") + .name = "binkaudio_rdft", + .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (RDFT)"), + .type = AVMEDIA_TYPE_AUDIO, + .id = AV_CODEC_ID_BINKAUDIO_RDFT, + .priv_data_size = sizeof(BinkAudioContext), + .init = decode_init, + .close = decode_end, + .decode = decode_frame, + .capabilities = CODEC_CAP_DELAY | CODEC_CAP_DR1, }; AVCodec ff_binkaudio_dct_decoder = { - "binkaudio_dct", - AVMEDIA_TYPE_AUDIO, - CODEC_ID_BINKAUDIO_DCT, - sizeof(BinkAudioContext), - decode_init, - NULL, - decode_end, - decode_frame, - .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (DCT)") + .name = "binkaudio_dct", + .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (DCT)"), + .type = AVMEDIA_TYPE_AUDIO, + .id = AV_CODEC_ID_BINKAUDIO_DCT, + .priv_data_size = sizeof(BinkAudioContext), + .init = decode_init, + .close = decode_end, + .decode = decode_frame, + .capabilities = CODEC_CAP_DELAY | CODEC_CAP_DR1, };