libavcodec/libvorbis.c
04d7f601
 /*
  * copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
  *
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
  */
 
983e3246
 /**
ba87f080
  * @file
983e3246
  * Ogg Vorbis codec support via libvorbisenc.
  * @author Mark Hills <mark@pogo.org.uk>
81e0d0b4
  */
 
 #include <vorbis/vorbisenc.h>
 
77336a5e
 #include "libavutil/opt.h"
81e0d0b4
 #include "avcodec.h"
2c124cb6
 #include "bytestream.h"
9577838f
 #include "vorbis.h"
954a6532
 #include "libavutil/mathematics.h"
81e0d0b4
 
3f4993f1
 #undef NDEBUG
 #include <assert.h>
 
6d8f985e
 #define OGGVORBIS_FRAME_SIZE 64
81e0d0b4
 
6d8f985e
 #define BUFFER_SIZE (1024*64)
81e0d0b4
 
 typedef struct OggVorbisContext {
77336a5e
     AVClass *av_class;
81e0d0b4
     vorbis_info vi ;
     vorbis_dsp_state vd ;
     vorbis_block vb ;
6d8f985e
     uint8_t buffer[BUFFER_SIZE];
     int buffer_index;
9c8f0768
     int eof;
3aca208a
 
     /* decoder */
     vorbis_comment vc ;
d76f581f
     ogg_packet op;
77336a5e
 
     double iblock;
81e0d0b4
 } OggVorbisContext ;
 
77336a5e
 static const AVOption options[]={
145f741e
 {"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},
77336a5e
 {NULL}
 };
da754858
 static const AVClass class = { "libvorbis", av_default_item_name, options, LIBAVUTIL_VERSION_INT };
81e0d0b4
 
5ef251e5
 static av_cold int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext) {
77336a5e
     OggVorbisContext *context = avccontext->priv_data ;
13c71451
     double cfreq;
c55427f8
 
1ff8f2b3
     if(avccontext->flags & CODEC_FLAG_QSCALE) {
13c71451
         /* variable bitrate */
         if(vorbis_encode_setup_vbr(vi, avccontext->channels,
1ff8f2b3
                 avccontext->sample_rate,
05276956
                 avccontext->global_quality / (float)FF_QP2LAMBDA / 10.0))
13c71451
             return -1;
     } else {
e5a5ea9e
         int minrate = avccontext->rc_min_rate > 0 ? avccontext->rc_min_rate : -1;
         int maxrate = avccontext->rc_min_rate > 0 ? avccontext->rc_max_rate : -1;
 
13c71451
         /* constant bitrate */
         if(vorbis_encode_setup_managed(vi, avccontext->channels,
e5a5ea9e
                 avccontext->sample_rate, minrate, avccontext->bit_rate, maxrate))
13c71451
             return -1;
 
57ebbccf
         /* variable bitrate by estimate, disable slow rate management */
         if(minrate == -1 && maxrate == -1)
             if(vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL))
                 return -1;
13c71451
     }
c55427f8
 
13c71451
     /* cutoff frequency */
     if(avccontext->cutoff > 0) {
         cfreq = avccontext->cutoff / 1000.0;
         if(vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))
             return -1;
     }
81e0d0b4
 
77336a5e
     if(context->iblock){
         vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &context->iblock);
     }
 
7c8c55ff
     if (avccontext->channels == 3 &&
             avccontext->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) ||
         avccontext->channels == 4 &&
             avccontext->channel_layout != AV_CH_LAYOUT_2_2 &&
             avccontext->channel_layout != AV_CH_LAYOUT_QUAD ||
         avccontext->channels == 5 &&
             avccontext->channel_layout != AV_CH_LAYOUT_5POINT0 &&
             avccontext->channel_layout != AV_CH_LAYOUT_5POINT0_BACK ||
         avccontext->channels == 6 &&
             avccontext->channel_layout != AV_CH_LAYOUT_5POINT1 &&
             avccontext->channel_layout != AV_CH_LAYOUT_5POINT1_BACK ||
         avccontext->channels == 7 &&
             avccontext->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) ||
         avccontext->channels == 8 &&
             avccontext->channel_layout != AV_CH_LAYOUT_7POINT1) {
         if (avccontext->channel_layout) {
             char name[32];
             av_get_channel_layout_string(name, sizeof(name), avccontext->channels,
                                          avccontext->channel_layout);
             av_log(avccontext, AV_LOG_ERROR, "%s not supported by Vorbis: "
                                              "output stream will have incorrect "
                                              "channel layout.\n", name);
         } else {
             av_log(avccontext, AV_LOG_WARNING, "No channel layout specified. The encoder "
                                                "will use Vorbis channel layout for "
                                                "%d channels.\n", avccontext->channels);
         }
     }
 
13c71451
     return vorbis_encode_setup_init(vi);
81e0d0b4
 }
 
fd7242dd
 /* How many bytes are needed for a buffer of length 'l' */
 static int xiph_len(int l) { return (1 + l / 255 + l); }
 
98a6fff9
 static av_cold int oggvorbis_encode_init(AVCodecContext *avccontext) {
81e0d0b4
     OggVorbisContext *context = avccontext->priv_data ;
bbb77e7c
     ogg_packet header, header_comm, header_code;
     uint8_t *p;
fd7242dd
     unsigned int offset;
81e0d0b4
 
     vorbis_info_init(&context->vi) ;
     if(oggvorbis_init_encoder(&context->vi, avccontext) < 0) {
fb0a835f
         av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed\n") ;
bb270c08
         return -1 ;
81e0d0b4
     }
     vorbis_analysis_init(&context->vd, &context->vi) ;
     vorbis_block_init(&context->vd, &context->vb) ;
 
bbb77e7c
     vorbis_comment_init(&context->vc);
     vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT) ;
 
     vorbis_analysis_headerout(&context->vd, &context->vc, &header,
                                 &header_comm, &header_code);
115329f1
 
fd7242dd
     avccontext->extradata_size=
         1 + xiph_len(header.bytes) + xiph_len(header_comm.bytes) +
         header_code.bytes;
     p = avccontext->extradata =
       av_malloc(avccontext->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
ad2b531d
     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);
     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;
fd7242dd
     assert(offset == avccontext->extradata_size);
115329f1
 
bbb77e7c
 /*    vorbis_block_clear(&context->vb);
     vorbis_dsp_clear(&context->vd);
     vorbis_info_clear(&context->vi);*/
     vorbis_comment_clear(&context->vc);
115329f1
 
81e0d0b4
     avccontext->frame_size = OGGVORBIS_FRAME_SIZE ;
115329f1
 
492cd3a9
     avccontext->coded_frame= avcodec_alloc_frame();
     avccontext->coded_frame->key_frame= 1;
115329f1
 
81e0d0b4
     return 0 ;
 }
 
 
3aca208a
 static int oggvorbis_encode_frame(AVCodecContext *avccontext,
bb270c08
                                   unsigned char *packets,
                            int buf_size, void *data)
81e0d0b4
 {
     OggVorbisContext *context = avccontext->priv_data ;
     ogg_packet op ;
0d00cf38
     signed short *audio = data ;
9c8f0768
     int l;
 
     if(data) {
42859ddb
         const int samples = avccontext->frame_size;
9c8f0768
         float **buffer ;
9577838f
         int c, channels = context->vi.channels;
81e0d0b4
 
a6080be7
         buffer = vorbis_analysis_buffer(&context->vd, samples) ;
9577838f
         for (c = 0; c < channels; c++) {
             int co = (channels > 8) ? c :
                 ff_vorbis_encoding_channel_layout_offsets[channels-1][c];
a6080be7
             for(l = 0 ; l < samples ; l++)
9577838f
                 buffer[c][l]=audio[l*channels+co]/32768.f;
bb270c08
         }
a6080be7
         vorbis_analysis_wrote(&context->vd, samples) ;
9c8f0768
     } else {
         if(!context->eof)
             vorbis_analysis_wrote(&context->vd, 0) ;
         context->eof = 1;
     }
81e0d0b4
 
     while(vorbis_analysis_blockout(&context->vd, &context->vb) == 1) {
bb270c08
         vorbis_analysis(&context->vb, NULL);
         vorbis_bitrate_addblock(&context->vb) ;
81e0d0b4
 
bb270c08
         while(vorbis_bitrate_flushpacket(&context->vd, &op)) {
0afd2a92
             /* 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. */
d99659f0
             if(op.bytes==1 && op.e_o_s)
6f824977
                 continue;
c426562c
             if (context->buffer_index + sizeof(ogg_packet) + op.bytes > BUFFER_SIZE) {
                 av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
                 return -1;
             }
6d8f985e
             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);
bb270c08
         }
81e0d0b4
     }
 
3f4993f1
     l=0;
6d8f985e
     if(context->buffer_index){
bbb77e7c
         ogg_packet *op2= (ogg_packet*)context->buffer;
6d8f985e
         op2->packet = context->buffer + sizeof(ogg_packet);
3f4993f1
 
6f824977
         l=  op2->bytes;
42661adf
         avccontext->coded_frame->pts= av_rescale_q(op2->granulepos, (AVRational){1, avccontext->sample_rate}, avccontext->time_base);
3723f01c
         //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
3f4993f1
 
c426562c
         if (l > buf_size) {
             av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
             return -1;
         }
 
6f824977
         memcpy(packets, op2->packet, l);
         context->buffer_index -= l + sizeof(ogg_packet);
1204a13c
         memmove(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
6d8f985e
 //        av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
     }
 
3f4993f1
     return l;
81e0d0b4
 }
 
 
98a6fff9
 static av_cold int oggvorbis_encode_close(AVCodecContext *avccontext) {
81e0d0b4
     OggVorbisContext *context = avccontext->priv_data ;
 /*  ogg_packet op ; */
115329f1
 
81e0d0b4
     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);
492cd3a9
 
     av_freep(&avccontext->coded_frame);
bbb77e7c
     av_freep(&avccontext->extradata);
115329f1
 
81e0d0b4
     return 0 ;
 }
 
 
e7e2df27
 AVCodec ff_libvorbis_encoder = {
86714887
     .name           = "libvorbis",
     .type           = AVMEDIA_TYPE_AUDIO,
     .id             = CODEC_ID_VORBIS,
     .priv_data_size = sizeof(OggVorbisContext),
     .init           = oggvorbis_encode_init,
     .encode         = oggvorbis_encode_frame,
     .close          = oggvorbis_encode_close,
     .capabilities   = CODEC_CAP_DELAY,
     .sample_fmts    = (const enum AVSampleFormat[]){AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE},
     .long_name      = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
     .priv_class     = &class,
 };