libavcodec/libvorbis.c

   1 /*
   2  * copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
   3  *
   4  * This file is part of Libav.
   5  *
   6  * Libav is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * Libav is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with Libav; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19  */
  20
  21 /**
  22  * @file
  23  * Ogg Vorbis codec support via libvorbisenc.
  24  * @author Mark Hills <mark@pogo.org.uk>
  25  */
  26
  27 #include <vorbis/vorbisenc.h>
  28
  29 #include "libavutil/opt.h"
  30 #include "avcodec.h"
  31 #include "bytestream.h"
  32 #include "vorbis.h"
  33 #include "libavutil/mathematics.h"
  34
  35 #undef NDEBUG
  36 #include <assert.h>
  37
  38 #define OGGVORBIS_FRAME_SIZE 64
  39
  40 #define BUFFER_SIZE (1024*64)
  41
  42 typedef struct OggVorbisContext {
  43     AVClass *av_class;
  44     vorbis_info vi ;
  45     vorbis_dsp_state vd ;
  46     vorbis_block vb ;
  47     uint8_t buffer[BUFFER_SIZE];
  48     int buffer_index;
  49     int eof;
  50
  51     /* decoder */
  52     vorbis_comment vc ;
  53     ogg_packet op;
  54
  55     double iblock;
  56 } OggVorbisContext ;
  57
  58 static const AVOption options[]={
  59 {"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},
  60 {NULL}
  61 };
  62 static const AVClass class = { "libvorbis", av_default_item_name, options, LIBAVUTIL_VERSION_INT };
  63
  64 static av_cold int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext) {
  65     OggVorbisContext *context = avccontext->priv_data ;
  66     double cfreq;
  67
  68     if(avccontext->flags & CODEC_FLAG_QSCALE) {
  69         /* variable bitrate */
  70         if(vorbis_encode_setup_vbr(vi, avccontext->channels,
  71                 avccontext->sample_rate,
  72                 avccontext->global_quality / (float)FF_QP2LAMBDA / 10.0))
  73             return -1;
  74     } else {
  75         int minrate = avccontext->rc_min_rate > 0 ? avccontext->rc_min_rate : -1;
  76         int maxrate = avccontext->rc_min_rate > 0 ? avccontext->rc_max_rate : -1;
  77
  78         /* constant bitrate */
  79         if(vorbis_encode_setup_managed(vi, avccontext->channels,
  80                 avccontext->sample_rate, minrate, avccontext->bit_rate, maxrate))
  81             return -1;
  82
  83         /* variable bitrate by estimate, disable slow rate management */
  84         if(minrate == -1 && maxrate == -1)
  85             if(vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL))
  86                 return -1;
  87     }
  88
  89     /* cutoff frequency */
  90     if(avccontext->cutoff > 0) {
  91         cfreq = avccontext->cutoff / 1000.0;
  92         if(vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))
  93             return -1;
  94     }
  95
  96     if(context->iblock){
  97         vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &context->iblock);
  98     }
  99
 100     return vorbis_encode_setup_init(vi);
 101 }
 102
 103 /* How many bytes are needed for a buffer of length 'l' */
 104 static int xiph_len(int l) { return (1 + l / 255 + l); }
 105
 106 static av_cold int oggvorbis_encode_init(AVCodecContext *avccontext) {
 107     OggVorbisContext *context = avccontext->priv_data ;
 108     ogg_packet header, header_comm, header_code;
 109     uint8_t *p;
 110     unsigned int offset;
 111
 112     vorbis_info_init(&context->vi) ;
 113     if(oggvorbis_init_encoder(&context->vi, avccontext) < 0) {
 114         av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed\n") ;
 115         return -1 ;
 116     }
 117     vorbis_analysis_init(&context->vd, &context->vi) ;
 118     vorbis_block_init(&context->vd, &context->vb) ;
 119
 120     vorbis_comment_init(&context->vc);
 121     vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT) ;
 122
 123     vorbis_analysis_headerout(&context->vd, &context->vc, &header,
 124                                 &header_comm, &header_code);
 125
 126     avccontext->extradata_size=
 127         1 + xiph_len(header.bytes) + xiph_len(header_comm.bytes) +
 128         header_code.bytes;
 129     p = avccontext->extradata =
 130       av_malloc(avccontext->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
 131     p[0] = 2;
 132     offset = 1;
 133     offset += av_xiphlacing(&p[offset], header.bytes);
 134     offset += av_xiphlacing(&p[offset], header_comm.bytes);
 135     memcpy(&p[offset], header.packet, header.bytes);
 136     offset += header.bytes;
 137     memcpy(&p[offset], header_comm.packet, header_comm.bytes);
 138     offset += header_comm.bytes;
 139     memcpy(&p[offset], header_code.packet, header_code.bytes);
 140     offset += header_code.bytes;
 141     assert(offset == avccontext->extradata_size);
 142
 143 /*    vorbis_block_clear(&context->vb);
 144     vorbis_dsp_clear(&context->vd);
 145     vorbis_info_clear(&context->vi);*/
 146     vorbis_comment_clear(&context->vc);
 147
 148     avccontext->frame_size = OGGVORBIS_FRAME_SIZE ;
 149
 150     avccontext->coded_frame= avcodec_alloc_frame();
 151     avccontext->coded_frame->key_frame= 1;
 152
 153     return 0 ;
 154 }
 155
 156
 157 static int oggvorbis_encode_frame(AVCodecContext *avccontext,
 158                                   unsigned char *packets,
 159                            int buf_size, void *data)
 160 {
 161     OggVorbisContext *context = avccontext->priv_data ;
 162     ogg_packet op ;
 163     signed short *audio = data ;
 164     int l;
 165
 166     if(data) {
 167         const int samples = avccontext->frame_size;
 168         float **buffer ;
 169         int c, channels = context->vi.channels;
 170
 171         buffer = vorbis_analysis_buffer(&context->vd, samples) ;
 172         for (c = 0; c < channels; c++) {
 173             int co = (channels > 8) ? c :
 174                 ff_vorbis_encoding_channel_layout_offsets[channels-1][c];
 175             for(l = 0 ; l < samples ; l++)
 176                 buffer[c][l]=audio[l*channels+co]/32768.f;
 177         }
 178         vorbis_analysis_wrote(&context->vd, samples) ;
 179     } else {
 180         if(!context->eof)
 181             vorbis_analysis_wrote(&context->vd, 0) ;
 182         context->eof = 1;
 183     }
 184
 185     while(vorbis_analysis_blockout(&context->vd, &context->vb) == 1) {
 186         vorbis_analysis(&context->vb, NULL);
 187         vorbis_bitrate_addblock(&context->vb) ;
 188
 189         while(vorbis_bitrate_flushpacket(&context->vd, &op)) {
 190             /* i'd love to say the following line is a hack, but sadly it's
 191              * not, apparently the end of stream decision is in libogg. */
 192             if(op.bytes==1 && op.e_o_s)
 193                 continue;
 194             if (context->buffer_index + sizeof(ogg_packet) + op.bytes > BUFFER_SIZE) {
 195                 av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
 196                 return -1;
 197             }
 198             memcpy(context->buffer + context->buffer_index, &op, sizeof(ogg_packet));
 199             context->buffer_index += sizeof(ogg_packet);
 200             memcpy(context->buffer + context->buffer_index, op.packet, op.bytes);
 201             context->buffer_index += op.bytes;
 202 //            av_log(avccontext, AV_LOG_DEBUG, "e%d / %d\n", context->buffer_index, op.bytes);
 203         }
 204     }
 205
 206     l=0;
 207     if(context->buffer_index){
 208         ogg_packet *op2= (ogg_packet*)context->buffer;
 209         op2->packet = context->buffer + sizeof(ogg_packet);
 210
 211         l=  op2->bytes;
 212         avccontext->coded_frame->pts= av_rescale_q(op2->granulepos, (AVRational){1, avccontext->sample_rate}, avccontext->time_base);
 213         //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
 214
 215         if (l > buf_size) {
 216             av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
 217             return -1;
 218         }
 219
 220         memcpy(packets, op2->packet, l);
 221         context->buffer_index -= l + sizeof(ogg_packet);
 222         memmove(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
 223 //        av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
 224     }
 225
 226     return l;
 227 }
 228
 229
 230 static av_cold int oggvorbis_encode_close(AVCodecContext *avccontext) {
 231     OggVorbisContext *context = avccontext->priv_data ;
 232 /*  ogg_packet op ; */
 233
 234     vorbis_analysis_wrote(&context->vd, 0) ; /* notify vorbisenc this is EOF */
 235
 236     vorbis_block_clear(&context->vb);
 237     vorbis_dsp_clear(&context->vd);
 238     vorbis_info_clear(&context->vi);
 239
 240     av_freep(&avccontext->coded_frame);
 241     av_freep(&avccontext->extradata);
 242
 243     return 0 ;
 244 }
 245
 246
 247 AVCodec ff_libvorbis_encoder = {
 248     .name           = "libvorbis",
 249     .type           = AVMEDIA_TYPE_AUDIO,
 250     .id             = CODEC_ID_VORBIS,
 251     .priv_data_size = sizeof(OggVorbisContext),
 252     .init           = oggvorbis_encode_init,
 253     .encode         = oggvorbis_encode_frame,
 254     .close          = oggvorbis_encode_close,
 255     .capabilities   = CODEC_CAP_DELAY,
 256     .sample_fmts    = (const enum AVSampleFormat[]){AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE},
 257     .long_name      = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
 258     .priv_class     = &class,
 259 };