libavcodec/libvorbisenc.c
04d7f601
 /*
c1d62489
  * Copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
04d7f601
  *
b78e7197
  * This file is part of FFmpeg.
  *
  * FFmpeg is free software; you can redistribute it and/or
04d7f601
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
b78e7197
  * version 2.1 of the License, or (at your option) any later version.
04d7f601
  *
b78e7197
  * FFmpeg is distributed in the hope that it will be useful,
04d7f601
  * 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
b78e7197
  * License along with FFmpeg; if not, write to the Free Software
e5a389a1
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
04d7f601
  */
 
81e0d0b4
 #include <vorbis/vorbisenc.h>
 
d014e195
 #include "libavutil/avassert.h"
592c4dbc
 #include "libavutil/fifo.h"
77336a5e
 #include "libavutil/opt.h"
81e0d0b4
 #include "avcodec.h"
e5aab2d7
 #include "audio_frame_queue.h"
91a28b0e
 #include "internal.h"
9577838f
 #include "vorbis.h"
e5aab2d7
 #include "vorbis_parser.h"
81e0d0b4
 
3f4993f1
 
eb35ef29
 /* 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.
  */
6d8f985e
 #define OGGVORBIS_FRAME_SIZE 64
81e0d0b4
 
ca5ab8cd
 #define BUFFER_SIZE (1024 * 64)
81e0d0b4
 
c1d62489
 typedef struct OggVorbisEncContext {
eb35ef29
     AVClass *av_class;                  /**< class for AVOptions            */
afcb6711
     AVFrame frame;
eb35ef29
     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 */
592c4dbc
     AVFifoBuffer *pkt_fifo;             /**< output packet buffer           */
eb35ef29
     int eof;                            /**< end-of-file flag               */
f15c4281
     int dsp_initialized;                /**< vd has been initialized        */
eb35ef29
     vorbis_comment vc;                  /**< VorbisComment info             */
     double iblock;                      /**< impulse block bias option      */
e5aab2d7
     VorbisParseContext vp;              /**< parse context to get durations */
     AudioFrameQueue afq;                /**< frame queue for timestamps     */
c1d62489
 } OggVorbisEncContext;
81e0d0b4
 
ca5ab8cd
 static const AVOption options[] = {
c1d62489
     { "iblock", "Sets the impulse block bias", offsetof(OggVorbisEncContext, iblock), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, -15, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
ca5ab8cd
     { NULL }
77336a5e
 };
147ff24a
 
 static const AVCodecDefault defaults[] = {
     { "b",  "0" },
     { NULL },
 };
 
e9d5c052
 static const AVClass class = {
     .class_name = "libvorbis",
     .item_name  = av_default_item_name,
     .option     = options,
     .version    = LIBAVUTIL_VERSION_INT,
 };
eb35ef29
 
6f600ab3
 static int vorbis_error_to_averror(int ov_err)
0098e79f
 {
6f600ab3
     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;
0098e79f
     }
 }
 
eb35ef29
 static av_cold int oggvorbis_init_encoder(vorbis_info *vi,
                                           AVCodecContext *avctx)
ca5ab8cd
 {
c1d62489
     OggVorbisEncContext *s = avctx->priv_data;
13c71451
     double cfreq;
6f600ab3
     int ret;
c55427f8
 
147ff24a
     if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) {
eb35ef29
         /* variable bitrate
          * NOTE: we use the oggenc range of -1 to 10 for global_quality for
147ff24a
          *       user convenience, but libvorbis uses -0.1 to 1.0.
eb35ef29
          */
         float q = avctx->global_quality / (float)FF_QP2LAMBDA;
147ff24a
         /* default to 3 if the user did not set quality or bitrate */
         if (!(avctx->flags & CODEC_FLAG_QSCALE))
             q = 3.0;
eb35ef29
         if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels,
                                            avctx->sample_rate,
6f600ab3
                                            q / 10.0)))
             goto error;
13c71451
     } else {
eb35ef29
         int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1;
182d4f1f
         int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1;
e5a5ea9e
 
eb35ef29
         /* average bitrate */
         if ((ret = vorbis_encode_setup_managed(vi, avctx->channels,
182d4f1f
                                                avctx->sample_rate, maxrate,
                                                avctx->bit_rate, minrate)))
6f600ab3
             goto error;
13c71451
 
57ebbccf
         /* variable bitrate by estimate, disable slow rate management */
ca5ab8cd
         if (minrate == -1 && maxrate == -1)
6f600ab3
             if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL)))
305e4b35
                 goto error; /* should not happen */
13c71451
     }
c55427f8
 
13c71451
     /* cutoff frequency */
eb35ef29
     if (avctx->cutoff > 0) {
         cfreq = avctx->cutoff / 1000.0;
6f600ab3
         if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
305e4b35
             goto error; /* should not happen */
13c71451
     }
81e0d0b4
 
eb35ef29
     /* impulse block bias */
     if (s->iblock) {
         if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock)))
6f600ab3
             goto error;
77336a5e
     }
 
79ae084e
     if (avctx->channels == 3 &&
             avctx->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
         avctx->channels == 4 &&
             avctx->channel_layout != AV_CH_LAYOUT_2_2 &&
             avctx->channel_layout != AV_CH_LAYOUT_QUAD ||
         avctx->channels == 5 &&
             avctx->channel_layout != AV_CH_LAYOUT_5POINT0 &&
             avctx->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
         avctx->channels == 6 &&
             avctx->channel_layout != AV_CH_LAYOUT_5POINT1 &&
             avctx->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
         avctx->channels == 7 &&
             avctx->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
         avctx->channels == 8 &&
             avctx->channel_layout != AV_CH_LAYOUT_7POINT1) {
         if (avctx->channel_layout) {
7c8c55ff
             char name[32];
79ae084e
             av_get_channel_layout_string(name, sizeof(name), avctx->channels,
                                          avctx->channel_layout);
             av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: "
7c8c55ff
                                              "output stream will have incorrect "
                                              "channel layout.\n", name);
         } else {
79ae084e
             av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder "
7c8c55ff
                                                "will use Vorbis channel layout for "
79ae084e
                                                "%d channels.\n", avctx->channels);
7c8c55ff
         }
     }
 
6f600ab3
     if ((ret = vorbis_encode_setup_init(vi)))
         goto error;
 
     return 0;
 error:
     return vorbis_error_to_averror(ret);
81e0d0b4
 }
 
fd7242dd
 /* How many bytes are needed for a buffer of length 'l' */
ca5ab8cd
 static int xiph_len(int l)
 {
d4b63054
     return 1 + l / 255 + l;
ca5ab8cd
 }
fd7242dd
 
eb35ef29
 static av_cold int oggvorbis_encode_close(AVCodecContext *avctx)
6f600ab3
 {
c1d62489
     OggVorbisEncContext *s = avctx->priv_data;
6f600ab3
 
eb35ef29
     /* notify vorbisenc this is EOF */
f15c4281
     if (s->dsp_initialized)
         vorbis_analysis_wrote(&s->vd, 0);
6f600ab3
 
eb35ef29
     vorbis_block_clear(&s->vb);
     vorbis_dsp_clear(&s->vd);
     vorbis_info_clear(&s->vi);
6f600ab3
 
592c4dbc
     av_fifo_free(s->pkt_fifo);
e5aab2d7
     ff_af_queue_close(&s->afq);
 #if FF_API_OLD_ENCODE_AUDIO
eb35ef29
     av_freep(&avctx->coded_frame);
e5aab2d7
 #endif
eb35ef29
     av_freep(&avctx->extradata);
6f600ab3
 
     return 0;
 }
 
eb35ef29
 static av_cold int oggvorbis_encode_init(AVCodecContext *avctx)
ca5ab8cd
 {
c1d62489
     OggVorbisEncContext *s = avctx->priv_data;
bbb77e7c
     ogg_packet header, header_comm, header_code;
     uint8_t *p;
fd7242dd
     unsigned int offset;
6f600ab3
     int ret;
81e0d0b4
 
eb35ef29
     vorbis_info_init(&s->vi);
     if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) {
a45a1ea5
         av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n");
6f600ab3
         goto error;
     }
eb35ef29
     if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) {
a45a1ea5
         av_log(avctx, AV_LOG_ERROR, "analysis init failed\n");
6f600ab3
         ret = vorbis_error_to_averror(ret);
         goto error;
     }
f15c4281
     s->dsp_initialized = 1;
eb35ef29
     if ((ret = vorbis_block_init(&s->vd, &s->vb))) {
a45a1ea5
         av_log(avctx, AV_LOG_ERROR, "dsp init failed\n");
6f600ab3
         ret = vorbis_error_to_averror(ret);
         goto error;
81e0d0b4
     }
 
eb35ef29
     vorbis_comment_init(&s->vc);
d9b92980
     if (!(avctx->flags & CODEC_FLAG_BITEXACT))
         vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT);
bbb77e7c
 
eb35ef29
     if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm,
                                          &header_code))) {
6f600ab3
         ret = vorbis_error_to_averror(ret);
         goto error;
     }
115329f1
 
eb35ef29
     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);
6f600ab3
     if (!p) {
         ret = AVERROR(ENOMEM);
         goto error;
     }
ca5ab8cd
     p[0]    = 2;
     offset  = 1;
ad2b531d
     offset += av_xiphlacing(&p[offset], header.bytes);
     offset += av_xiphlacing(&p[offset], header_comm.bytes);
     memcpy(&p[offset], header.packet, header.bytes);
     offset += header.bytes;
     memcpy(&p[offset], header_comm.packet, header_comm.bytes);
     offset += header_comm.bytes;
     memcpy(&p[offset], header_code.packet, header_code.bytes);
     offset += header_code.bytes;
d014e195
     av_assert0(offset == avctx->extradata_size);
115329f1
 
e5aab2d7
     if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) {
         av_log(avctx, AV_LOG_ERROR, "invalid extradata\n");
         return ret;
     }
 
eb35ef29
     vorbis_comment_clear(&s->vc);
115329f1
 
eb35ef29
     avctx->frame_size = OGGVORBIS_FRAME_SIZE;
e5aab2d7
     ff_af_queue_init(avctx, &s->afq);
115329f1
 
592c4dbc
     s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE);
     if (!s->pkt_fifo) {
         ret = AVERROR(ENOMEM);
         goto error;
     }
115329f1
 
e5aab2d7
 #if FF_API_OLD_ENCODE_AUDIO
eb35ef29
     avctx->coded_frame = avcodec_alloc_frame();
     if (!avctx->coded_frame) {
6f600ab3
         ret = AVERROR(ENOMEM);
         goto error;
     }
e5aab2d7
 #endif
115329f1
 
ca5ab8cd
     return 0;
6f600ab3
 error:
eb35ef29
     oggvorbis_encode_close(avctx);
6f600ab3
     return ret;
81e0d0b4
 }
 
e5aab2d7
 static int oggvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
                                   const AVFrame *frame, int *got_packet_ptr)
81e0d0b4
 {
c1d62489
     OggVorbisEncContext *s = avctx->priv_data;
ca5ab8cd
     ogg_packet op;
e5aab2d7
     int ret, duration;
9c8f0768
 
eb35ef29
     /* send samples to libvorbis */
e5aab2d7
     if (frame) {
         const int samples = frame->nb_samples;
ca5ab8cd
         float **buffer;
eb35ef29
         int c, channels = s->vi.channels;
81e0d0b4
 
eb35ef29
         buffer = vorbis_analysis_buffer(&s->vd, samples);
9577838f
         for (c = 0; c < channels; c++) {
             int co = (channels > 8) ? c :
ca5ab8cd
                      ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
cfc0a80a
             memcpy(buffer[c], frame->extended_data[co],
                    samples * sizeof(*buffer[c]));
bb270c08
         }
a45a1ea5
         if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) {
             av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
94025d8a
             return vorbis_error_to_averror(ret);
bb270c08
         }
3be8aeb1
         if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
e5aab2d7
             return ret;
9c8f0768
     } else {
eb35ef29
         if (!s->eof)
a45a1ea5
             if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) {
                 av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n");
94025d8a
                 return vorbis_error_to_averror(ret);
a45a1ea5
             }
eb35ef29
         s->eof = 1;
9c8f0768
     }
81e0d0b4
 
eb35ef29
     /* retrieve available packets from libvorbis */
94025d8a
     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;
81e0d0b4
 
eb35ef29
         /* add any available packets to the output packet buffer */
94025d8a
         while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) {
592c4dbc
             if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) {
79ae084e
                 av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n");
a45a1ea5
                 return AVERROR_BUG;
c426562c
             }
592c4dbc
             av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
             av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL);
bb270c08
         }
a45a1ea5
         if (ret < 0) {
             av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
94025d8a
             break;
bb270c08
         }
81e0d0b4
     }
a45a1ea5
     if (ret < 0) {
         av_log(avctx, AV_LOG_ERROR, "error getting available packets\n");
94025d8a
         return vorbis_error_to_averror(ret);
a45a1ea5
     }
81e0d0b4
 
e5aab2d7
     /* check for available packets */
     if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet))
         return 0;
 
     av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL);
 
ae2c33b0
     if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes)))
e5aab2d7
         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 */
8c414534
         if (!avctx->delay && s->afq.frames) {
e5aab2d7
             avctx->delay              = duration;
2c34367b
             av_assert0(!s->afq.remaining_delay);
             s->afq.frames->duration  += duration;
bd226697
             if (s->afq.frames->pts != AV_NOPTS_VALUE)
                 s->afq.frames->pts       -= duration;
e5aab2d7
             s->afq.remaining_samples += duration;
c426562c
         }
e5aab2d7
         ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration);
6d8f985e
     }
 
e5aab2d7
     *got_packet_ptr = 1;
     return 0;
81e0d0b4
 }
 
e7e2df27
 AVCodec ff_libvorbis_encoder = {
86714887
     .name           = "libvorbis",
     .type           = AVMEDIA_TYPE_AUDIO,
36ef5369
     .id             = AV_CODEC_ID_VORBIS,
c1d62489
     .priv_data_size = sizeof(OggVorbisEncContext),
86714887
     .init           = oggvorbis_encode_init,
e5aab2d7
     .encode2        = oggvorbis_encode_frame,
86714887
     .close          = oggvorbis_encode_close,
     .capabilities   = CODEC_CAP_DELAY,
cfc0a80a
     .sample_fmts    = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
eb35ef29
                                                       AV_SAMPLE_FMT_NONE },
c1d62489
     .long_name      = NULL_IF_CONFIG_SMALL("libvorbis"),
86714887
     .priv_class     = &class,
147ff24a
     .defaults       = defaults,
86714887
 };