X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Flibvorbis.c;h=1d7b7ef49b8edc3db3fee94aed09f615e8bd7e7d;hb=0b1c868508ba092c63bfe429fe57ec1379afa502;hp=faaceb0ba7621d6ffbb039405873d6feafd7b808;hpb=aee481cebe8f95ce3789bdead6fb8ddfb142c37f;p=ffmpeg diff --git a/libavcodec/libvorbis.c b/libavcodec/libvorbis.c index faaceb0ba76..1d7b7ef49b8 100644 --- a/libavcodec/libvorbis.c +++ b/libavcodec/libvorbis.c @@ -1,111 +1,220 @@ /* * copyright (c) 2002 Mark Hills * - * This file is part of FFmpeg. + * This file is part of Libav. * - * FFmpeg is free software; you can redistribute it and/or + * Libav 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, + * Libav 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 + * License along with Libav; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ /** - * @file oggvorbis.c - * Ogg Vorbis codec support via libvorbisenc. + * @file + * Vorbis encoding support via libvorbisenc. * @author Mark Hills */ #include +#include "libavutil/fifo.h" +#include "libavutil/opt.h" #include "avcodec.h" +#include "audio_frame_queue.h" #include "bytestream.h" +#include "internal.h" +#include "vorbis.h" +#include "vorbis_parser.h" #undef NDEBUG #include +/* Number of samples the user should send in each call. + * This value is used because it is the LCD of all possible frame sizes, so + * an output packet will always start at the same point as one of the input + * packets. + */ #define OGGVORBIS_FRAME_SIZE 64 -#define BUFFER_SIZE (1024*64) +#define BUFFER_SIZE (1024 * 64) typedef struct OggVorbisContext { - vorbis_info vi ; - vorbis_dsp_state vd ; - vorbis_block vb ; - uint8_t buffer[BUFFER_SIZE]; - int buffer_index; - - /* decoder */ - vorbis_comment vc ; - ogg_packet op; -} OggVorbisContext ; - + AVClass *av_class; /**< class for AVOptions */ + vorbis_info vi; /**< vorbis_info used during init */ + vorbis_dsp_state vd; /**< DSP state used for analysis */ + vorbis_block vb; /**< vorbis_block used for analysis */ + AVFifoBuffer *pkt_fifo; /**< output packet buffer */ + int eof; /**< end-of-file flag */ + int dsp_initialized; /**< vd has been initialized */ + vorbis_comment vc; /**< VorbisComment info */ + ogg_packet op; /**< ogg packet */ + double iblock; /**< impulse block bias option */ + VorbisParseContext vp; /**< parse context to get durations */ + AudioFrameQueue afq; /**< frame queue for timestamps */ +} OggVorbisContext; + +static const AVOption options[] = { + { "iblock", "Sets the impulse block bias", offsetof(OggVorbisContext, iblock), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, -15, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM }, + { NULL } +}; + +static const AVCodecDefault defaults[] = { + { "b", "0" }, + { NULL }, +}; + +static const AVClass class = { "libvorbis", av_default_item_name, options, LIBAVUTIL_VERSION_INT }; + + +static int vorbis_error_to_averror(int ov_err) +{ + switch (ov_err) { + case OV_EFAULT: return AVERROR_BUG; + case OV_EINVAL: return AVERROR(EINVAL); + case OV_EIMPL: return AVERROR(EINVAL); + default: return AVERROR_UNKNOWN; + } +} -static int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext) { +static av_cold int oggvorbis_init_encoder(vorbis_info *vi, + AVCodecContext *avctx) +{ + OggVorbisContext *s = avctx->priv_data; double cfreq; - - if(avccontext->flags & CODEC_FLAG_QSCALE) { - /* variable bitrate */ - if(vorbis_encode_setup_vbr(vi, avccontext->channels, - avccontext->sample_rate, - avccontext->global_quality / (float)FF_QP2LAMBDA)) - return -1; + int ret; + + if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) { + /* variable bitrate + * NOTE: we use the oggenc range of -1 to 10 for global_quality for + * user convenience, but libvorbis uses -0.1 to 1.0. + */ + float q = avctx->global_quality / (float)FF_QP2LAMBDA; + /* default to 3 if the user did not set quality or bitrate */ + if (!(avctx->flags & CODEC_FLAG_QSCALE)) + q = 3.0; + if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels, + avctx->sample_rate, + q / 10.0))) + goto error; } else { - /* constant bitrate */ - if(vorbis_encode_setup_managed(vi, avccontext->channels, - avccontext->sample_rate, -1, avccontext->bit_rate, -1)) - return -1; - -#ifdef OGGVORBIS_VBR_BY_ESTIMATE - /* variable bitrate by estimate */ - if(vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE_AVG, NULL)) - return -1; -#endif + int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1; + int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1; + + /* average bitrate */ + if ((ret = vorbis_encode_setup_managed(vi, avctx->channels, + avctx->sample_rate, maxrate, + avctx->bit_rate, minrate))) + goto error; + + /* variable bitrate by estimate, disable slow rate management */ + if (minrate == -1 && maxrate == -1) + if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL))) + goto error; } /* cutoff frequency */ - if(avccontext->cutoff > 0) { - cfreq = avccontext->cutoff / 1000.0; - if(vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)) - return -1; + if (avctx->cutoff > 0) { + cfreq = avctx->cutoff / 1000.0; + if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))) + goto error; } - return vorbis_encode_setup_init(vi); + /* impulse block bias */ + if (s->iblock) { + if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock))) + goto error; + } + + if ((ret = vorbis_encode_setup_init(vi))) + goto error; + + return 0; +error: + return vorbis_error_to_averror(ret); +} + +/* How many bytes are needed for a buffer of length 'l' */ +static int xiph_len(int l) +{ + return 1 + l / 255 + l; +} + +static av_cold int oggvorbis_encode_close(AVCodecContext *avctx) +{ + OggVorbisContext *s = avctx->priv_data; + + /* notify vorbisenc this is EOF */ + if (s->dsp_initialized) + vorbis_analysis_wrote(&s->vd, 0); + + vorbis_block_clear(&s->vb); + vorbis_dsp_clear(&s->vd); + vorbis_info_clear(&s->vi); + + av_fifo_free(s->pkt_fifo); + ff_af_queue_close(&s->afq); +#if FF_API_OLD_ENCODE_AUDIO + av_freep(&avctx->coded_frame); +#endif + av_freep(&avctx->extradata); + + return 0; } -static int oggvorbis_encode_init(AVCodecContext *avccontext) { - OggVorbisContext *context = avccontext->priv_data ; +static av_cold int oggvorbis_encode_init(AVCodecContext *avctx) +{ + OggVorbisContext *s = avctx->priv_data; ogg_packet header, header_comm, header_code; uint8_t *p; - unsigned int offset, len; + unsigned int offset; + int ret; - vorbis_info_init(&context->vi) ; - if(oggvorbis_init_encoder(&context->vi, avccontext) < 0) { - av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed") ; - return -1 ; + vorbis_info_init(&s->vi); + if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) { + av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n"); + goto error; + } + if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) { + av_log(avctx, AV_LOG_ERROR, "analysis init failed\n"); + ret = vorbis_error_to_averror(ret); + goto error; + } + s->dsp_initialized = 1; + if ((ret = vorbis_block_init(&s->vd, &s->vb))) { + av_log(avctx, AV_LOG_ERROR, "dsp init failed\n"); + ret = vorbis_error_to_averror(ret); + goto error; } - vorbis_analysis_init(&context->vd, &context->vi) ; - vorbis_block_init(&context->vd, &context->vb) ; - vorbis_comment_init(&context->vc); - vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT) ; + vorbis_comment_init(&s->vc); + vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT); - vorbis_analysis_headerout(&context->vd, &context->vc, &header, - &header_comm, &header_code); + if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm, + &header_code))) { + ret = vorbis_error_to_averror(ret); + goto error; + } - len = header.bytes + header_comm.bytes + header_code.bytes; - avccontext->extradata_size= 64 + len + len/255; - p = avccontext->extradata= av_mallocz(avccontext->extradata_size); - p[0] = 2; - offset = 1; + avctx->extradata_size = 1 + xiph_len(header.bytes) + + xiph_len(header_comm.bytes) + + header_code.bytes; + p = avctx->extradata = av_malloc(avctx->extradata_size + + FF_INPUT_BUFFER_PADDING_SIZE); + if (!p) { + ret = AVERROR(ENOMEM); + goto error; + } + p[0] = 2; + offset = 1; offset += av_xiphlacing(&p[offset], header.bytes); offset += av_xiphlacing(&p[offset], header_comm.bytes); memcpy(&p[offset], header.packet, header.bytes); @@ -114,269 +223,143 @@ static int oggvorbis_encode_init(AVCodecContext *avccontext) { offset += header_comm.bytes; memcpy(&p[offset], header_code.packet, header_code.bytes); offset += header_code.bytes; - avccontext->extradata_size = offset; - avccontext->extradata= av_realloc(avccontext->extradata, avccontext->extradata_size); - -/* vorbis_block_clear(&context->vb); - vorbis_dsp_clear(&context->vd); - vorbis_info_clear(&context->vi);*/ - vorbis_comment_clear(&context->vc); + assert(offset == avctx->extradata_size); - avccontext->frame_size = OGGVORBIS_FRAME_SIZE ; - - avccontext->coded_frame= avcodec_alloc_frame(); - avccontext->coded_frame->key_frame= 1; - - return 0 ; -} - - -static int oggvorbis_encode_frame(AVCodecContext *avccontext, - unsigned char *packets, - int buf_size, void *data) -{ - OggVorbisContext *context = avccontext->priv_data ; - float **buffer ; - ogg_packet op ; - signed short *audio = data ; - int l, samples = data ? OGGVORBIS_FRAME_SIZE : 0; - - buffer = vorbis_analysis_buffer(&context->vd, samples) ; - - if(context->vi.channels == 1) { - for(l = 0 ; l < samples ; l++) - buffer[0][l]=audio[l]/32768.f; - } else { - for(l = 0 ; l < samples ; l++){ - buffer[0][l]=audio[l*2]/32768.f; - buffer[1][l]=audio[l*2+1]/32768.f; - } + if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) { + av_log(avctx, AV_LOG_ERROR, "invalid extradata\n"); + return ret; } - vorbis_analysis_wrote(&context->vd, samples) ; - - while(vorbis_analysis_blockout(&context->vd, &context->vb) == 1) { - vorbis_analysis(&context->vb, NULL); - vorbis_bitrate_addblock(&context->vb) ; - - while(vorbis_bitrate_flushpacket(&context->vd, &op)) { - /* i'd love to say the following line is a hack, but sadly it's - * not, apparently the end of stream decision is in libogg. */ - if(op.bytes==1) - continue; - memcpy(context->buffer + context->buffer_index, &op, sizeof(ogg_packet)); - context->buffer_index += sizeof(ogg_packet); - memcpy(context->buffer + context->buffer_index, op.packet, op.bytes); - context->buffer_index += op.bytes; -// av_log(avccontext, AV_LOG_DEBUG, "e%d / %d\n", context->buffer_index, op.bytes); - } - } + vorbis_comment_clear(&s->vc); - l=0; - if(context->buffer_index){ - ogg_packet *op2= (ogg_packet*)context->buffer; - op2->packet = context->buffer + sizeof(ogg_packet); + avctx->frame_size = OGGVORBIS_FRAME_SIZE; + ff_af_queue_init(avctx, &s->afq); - l= op2->bytes; - avccontext->coded_frame->pts= av_rescale_q(op2->granulepos, (AVRational){1, avccontext->sample_rate}, avccontext->time_base); - //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate - - memcpy(packets, op2->packet, l); - context->buffer_index -= l + sizeof(ogg_packet); - memcpy(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index); -// av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l); + s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE); + if (!s->pkt_fifo) { + ret = AVERROR(ENOMEM); + goto error; } - return l; -} - - -static int oggvorbis_encode_close(AVCodecContext *avccontext) { - OggVorbisContext *context = avccontext->priv_data ; -/* ogg_packet op ; */ - - vorbis_analysis_wrote(&context->vd, 0) ; /* notify vorbisenc this is EOF */ - - vorbis_block_clear(&context->vb); - vorbis_dsp_clear(&context->vd); - vorbis_info_clear(&context->vi); - - av_freep(&avccontext->coded_frame); - av_freep(&avccontext->extradata); +#if FF_API_OLD_ENCODE_AUDIO + avctx->coded_frame = avcodec_alloc_frame(); + if (!avctx->coded_frame) { + ret = AVERROR(ENOMEM); + goto error; + } +#endif - return 0 ; + return 0; +error: + oggvorbis_encode_close(avctx); + return ret; } - -AVCodec libvorbis_encoder = { - "libvorbis", - CODEC_TYPE_AUDIO, - CODEC_ID_VORBIS, - sizeof(OggVorbisContext), - oggvorbis_encode_init, - oggvorbis_encode_frame, - oggvorbis_encode_close, - .capabilities= CODEC_CAP_DELAY, -} ; - -static int oggvorbis_decode_init(AVCodecContext *avccontext) { - OggVorbisContext *context = avccontext->priv_data ; - uint8_t *p= avccontext->extradata; - int i, hsizes[3]; - unsigned char *headers[3], *extradata = avccontext->extradata; - - vorbis_info_init(&context->vi) ; - vorbis_comment_init(&context->vc) ; - - if(! avccontext->extradata_size || ! p) { - av_log(avccontext, AV_LOG_ERROR, "vorbis extradata absent\n"); - return -1; - } - - if(p[0] == 0 && p[1] == 30) { - for(i = 0; i < 3; i++){ - hsizes[i] = bytestream_get_be16(&p); - headers[i] = p; - p += hsizes[i]; +static int oggvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, + const AVFrame *frame, int *got_packet_ptr) +{ + OggVorbisContext *s = avctx->priv_data; + ogg_packet op; + int ret, duration; + + /* send samples to libvorbis */ + if (frame) { + const float *audio = (const float *)frame->data[0]; + const int samples = frame->nb_samples; + float **buffer; + int c, channels = s->vi.channels; + + buffer = vorbis_analysis_buffer(&s->vd, samples); + for (c = 0; c < channels; c++) { + int i; + int co = (channels > 8) ? c : + ff_vorbis_encoding_channel_layout_offsets[channels - 1][c]; + for (i = 0; i < samples; i++) + buffer[c][i] = audio[i * channels + co]; } - } else if(*p == 2) { - unsigned int offset = 1; - p++; - for(i=0; i<2; i++) { - hsizes[i] = 0; - while((*p == 0xFF) && (offset < avccontext->extradata_size)) { - hsizes[i] += 0xFF; - offset++; - p++; - } - if(offset >= avccontext->extradata_size - 1) { - av_log(avccontext, AV_LOG_ERROR, - "vorbis header sizes damaged\n"); - return -1; - } - hsizes[i] += *p; - offset++; - p++; + if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) { + av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n"); + return vorbis_error_to_averror(ret); } - hsizes[2] = avccontext->extradata_size - hsizes[0]-hsizes[1]-offset; -#if 0 - av_log(avccontext, AV_LOG_DEBUG, - "vorbis header sizes: %d, %d, %d, / extradata_len is %d \n", - hsizes[0], hsizes[1], hsizes[2], avccontext->extradata_size); -#endif - headers[0] = extradata + offset; - headers[1] = extradata + offset + hsizes[0]; - headers[2] = extradata + offset + hsizes[0] + hsizes[1]; + if ((ret = ff_af_queue_add(&s->afq, frame) < 0)) + return ret; } else { - av_log(avccontext, AV_LOG_ERROR, - "vorbis initial header len is wrong: %d\n", *p); - return -1; + if (!s->eof) + if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) { + av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n"); + return vorbis_error_to_averror(ret); + } + s->eof = 1; } - for(i=0; i<3; i++){ - context->op.b_o_s= i==0; - context->op.bytes = hsizes[i]; - context->op.packet = headers[i]; - if(vorbis_synthesis_headerin(&context->vi, &context->vc, &context->op)<0){ - av_log(avccontext, AV_LOG_ERROR, "%d. vorbis header damaged\n", i+1); - return -1; + /* retrieve available packets from libvorbis */ + while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) { + if ((ret = vorbis_analysis(&s->vb, NULL)) < 0) + break; + if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0) + break; + + /* add any available packets to the output packet buffer */ + while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) { + if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) { + av_log(avctx, AV_LOG_ERROR, "packet buffer is too small"); + return AVERROR_BUG; + } + av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL); + av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL); } - } - - avccontext->channels = context->vi.channels; - avccontext->sample_rate = context->vi.rate; - avccontext->time_base= (AVRational){1, avccontext->sample_rate}; - - vorbis_synthesis_init(&context->vd, &context->vi); - vorbis_block_init(&context->vd, &context->vb); - - return 0 ; -} - - -static inline int conv(int samples, float **pcm, char *buf, int channels) { - int i, j, val ; - ogg_int16_t *ptr, *data = (ogg_int16_t*)buf ; - float *mono ; - - for(i = 0 ; i < channels ; i++){ - ptr = &data[i]; - mono = pcm[i] ; - - for(j = 0 ; j < samples ; j++) { - - val = mono[j] * 32767.f; - - val = av_clip_int16(val); - - *ptr = val ; - ptr += channels; + if (ret < 0) { + av_log(avctx, AV_LOG_ERROR, "error getting available packets\n"); + break; } } - - return 0 ; -} - - -static int oggvorbis_decode_frame(AVCodecContext *avccontext, - void *data, int *data_size, - uint8_t *buf, int buf_size) -{ - OggVorbisContext *context = avccontext->priv_data ; - float **pcm ; - ogg_packet *op= &context->op; - int samples, total_samples, total_bytes; - - if(!buf_size){ - //FIXME flush - return 0; + if (ret < 0) { + av_log(avctx, AV_LOG_ERROR, "error getting available packets\n"); + return vorbis_error_to_averror(ret); } - op->packet = buf; - op->bytes = buf_size; - -// av_log(avccontext, AV_LOG_DEBUG, "%d %d %d %"PRId64" %"PRId64" %d %d\n", op->bytes, op->b_o_s, op->e_o_s, op->granulepos, op->packetno, buf_size, context->vi.rate); - -/* for(i=0; ibytes; i++) - av_log(avccontext, AV_LOG_DEBUG, "%02X ", op->packet[i]); - av_log(avccontext, AV_LOG_DEBUG, "\n");*/ - - if(vorbis_synthesis(&context->vb, op) == 0) - vorbis_synthesis_blockin(&context->vd, &context->vb) ; + /* check for available packets */ + if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet)) + return 0; - total_samples = 0 ; - total_bytes = 0 ; + av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL); - while((samples = vorbis_synthesis_pcmout(&context->vd, &pcm)) > 0) { - conv(samples, pcm, (char*)data + total_bytes, context->vi.channels) ; - total_bytes += samples * 2 * context->vi.channels ; - total_samples += samples ; - vorbis_synthesis_read(&context->vd, samples) ; + if ((ret = ff_alloc_packet(avpkt, op.bytes))) { + av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n"); + return ret; + } + av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL); + + avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos); + + duration = avpriv_vorbis_parse_frame(&s->vp, avpkt->data, avpkt->size); + if (duration > 0) { + /* we do not know encoder delay until we get the first packet from + * libvorbis, so we have to update the AudioFrameQueue counts */ + if (!avctx->delay) { + avctx->delay = duration; + s->afq.remaining_delay += duration; + s->afq.remaining_samples += duration; + } + ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration); } - *data_size = total_bytes ; - return buf_size ; -} - - -static int oggvorbis_decode_close(AVCodecContext *avccontext) { - OggVorbisContext *context = avccontext->priv_data ; - - vorbis_info_clear(&context->vi) ; - vorbis_comment_clear(&context->vc) ; - - return 0 ; + *got_packet_ptr = 1; + return 0; } - -AVCodec libvorbis_decoder = { - "libvorbis", - CODEC_TYPE_AUDIO, - CODEC_ID_VORBIS, - sizeof(OggVorbisContext), - oggvorbis_decode_init, - NULL, - oggvorbis_decode_close, - oggvorbis_decode_frame, - .capabilities= CODEC_CAP_DELAY, -} ; +AVCodec ff_libvorbis_encoder = { + .name = "libvorbis", + .type = AVMEDIA_TYPE_AUDIO, + .id = CODEC_ID_VORBIS, + .priv_data_size = sizeof(OggVorbisContext), + .init = oggvorbis_encode_init, + .encode2 = oggvorbis_encode_frame, + .close = oggvorbis_encode_close, + .capabilities = CODEC_CAP_DELAY, + .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLT, + AV_SAMPLE_FMT_NONE }, + .long_name = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"), + .priv_class = &class, + .defaults = defaults, +};