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 {
  66     OggVorbisContext *context = avccontext->priv_data;
  67     double cfreq;
  68
  69     if (avccontext->flags & CODEC_FLAG_QSCALE) {
  70         /* variable bitrate */
  71         if (vorbis_encode_setup_vbr(vi, avccontext->channels,
  72                                     avccontext->sample_rate,
  73                                     avccontext->global_quality / (float)FF_QP2LAMBDA / 10.0))
  74             return -1;
  75     } else {
  76         int minrate = avccontext->rc_min_rate > 0 ? avccontext->rc_min_rate : -1;
  77         int maxrate = avccontext->rc_min_rate > 0 ? avccontext->rc_max_rate : -1;
  78
  79         /* constant bitrate */
  80         if (vorbis_encode_setup_managed(vi, avccontext->channels,
  81                                         avccontext->sample_rate, minrate,
  82                                         avccontext->bit_rate, maxrate))
  83             return -1;
  84
  85         /* variable bitrate by estimate, disable slow rate management */
  86         if (minrate == -1 && maxrate == -1)
  87             if (vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL))
  88                 return -1;
  89     }
  90
  91     /* cutoff frequency */
  92     if (avccontext->cutoff > 0) {
  93         cfreq = avccontext->cutoff / 1000.0;
  94         if (vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))
  95             return -1;
  96     }
  97
  98     if (context->iblock) {
  99         vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &context->iblock);
 100     }
 101
 102     return vorbis_encode_setup_init(vi);
 103 }
 104
 105 /* How many bytes are needed for a buffer of length 'l' */
 106 static int xiph_len(int l)
 107 {
 108     return (1 + l / 255 + l);
 109 }
 110
 111 static av_cold int oggvorbis_encode_init(AVCodecContext *avccontext)
 112 {
 113     OggVorbisContext *context = avccontext->priv_data;
 114     ogg_packet header, header_comm, header_code;
 115     uint8_t *p;
 116     unsigned int offset;
 117
 118     vorbis_info_init(&context->vi);
 119     if (oggvorbis_init_encoder(&context->vi, avccontext) < 0) {
 120         av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed\n");
 121         return -1;
 122     }
 123     vorbis_analysis_init(&context->vd, &context->vi);
 124     vorbis_block_init(&context->vd, &context->vb);
 125
 126     vorbis_comment_init(&context->vc);
 127     vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT);
 128
 129     vorbis_analysis_headerout(&context->vd, &context->vc, &header,
 130                               &header_comm, &header_code);
 131
 132     avccontext->extradata_size =
 133         1 + xiph_len(header.bytes) + xiph_len(header_comm.bytes) +
 134         header_code.bytes;
 135     p = avccontext->extradata =
 136             av_malloc(avccontext->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
 137     p[0]    = 2;
 138     offset  = 1;
 139     offset += av_xiphlacing(&p[offset], header.bytes);
 140     offset += av_xiphlacing(&p[offset], header_comm.bytes);
 141     memcpy(&p[offset], header.packet, header.bytes);
 142     offset += header.bytes;
 143     memcpy(&p[offset], header_comm.packet, header_comm.bytes);
 144     offset += header_comm.bytes;
 145     memcpy(&p[offset], header_code.packet, header_code.bytes);
 146     offset += header_code.bytes;
 147     assert(offset == avccontext->extradata_size);
 148
 149 #if 0
 150     vorbis_block_clear(&context->vb);
 151     vorbis_dsp_clear(&context->vd);
 152     vorbis_info_clear(&context->vi);
 153 #endif
 154     vorbis_comment_clear(&context->vc);
 155
 156     avccontext->frame_size = OGGVORBIS_FRAME_SIZE;
 157
 158     avccontext->coded_frame = avcodec_alloc_frame();
 159     avccontext->coded_frame->key_frame = 1;
 160
 161     return 0;
 162 }
 163
 164 static int oggvorbis_encode_frame(AVCodecContext *avccontext,
 165                                   unsigned char *packets,
 166                                   int buf_size, void *data)
 167 {
 168     OggVorbisContext *context = avccontext->priv_data;
 169     ogg_packet op;
 170     signed short *audio = data;
 171     int l;
 172
 173     if (data) {
 174         const int samples = avccontext->frame_size;
 175         float **buffer;
 176         int c, channels = context->vi.channels;
 177
 178         buffer = vorbis_analysis_buffer(&context->vd, samples);
 179         for (c = 0; c < channels; c++) {
 180             int co = (channels > 8) ? c :
 181                      ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
 182             for (l = 0; l < samples; l++)
 183                 buffer[c][l] = audio[l * channels + co] / 32768.f;
 184         }
 185         vorbis_analysis_wrote(&context->vd, samples);
 186     } else {
 187         if (!context->eof)
 188             vorbis_analysis_wrote(&context->vd, 0);
 189         context->eof = 1;
 190     }
 191
 192     while (vorbis_analysis_blockout(&context->vd, &context->vb) == 1) {
 193         vorbis_analysis(&context->vb, NULL);
 194         vorbis_bitrate_addblock(&context->vb);
 195
 196         while (vorbis_bitrate_flushpacket(&context->vd, &op)) {
 197             /* i'd love to say the following line is a hack, but sadly it's
 198              * not, apparently the end of stream decision is in libogg. */
 199             if (op.bytes == 1 && op.e_o_s)
 200                 continue;
 201             if (context->buffer_index + sizeof(ogg_packet) + op.bytes > BUFFER_SIZE) {
 202                 av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
 203                 return -1;
 204             }
 205             memcpy(context->buffer + context->buffer_index, &op, sizeof(ogg_packet));
 206             context->buffer_index += sizeof(ogg_packet);
 207             memcpy(context->buffer + context->buffer_index, op.packet, op.bytes);
 208             context->buffer_index += op.bytes;
 209 //            av_log(avccontext, AV_LOG_DEBUG, "e%d / %d\n", context->buffer_index, op.bytes);
 210         }
 211     }
 212
 213     l = 0;
 214     if (context->buffer_index) {
 215         ogg_packet *op2 = (ogg_packet *)context->buffer;
 216         op2->packet = context->buffer + sizeof(ogg_packet);
 217
 218         l = op2->bytes;
 219         avccontext->coded_frame->pts = av_rescale_q(op2->granulepos, (AVRational) { 1, avccontext->sample_rate }, avccontext->time_base);
 220         //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
 221
 222         if (l > buf_size) {
 223             av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.");
 224             return -1;
 225         }
 226
 227         memcpy(packets, op2->packet, l);
 228         context->buffer_index -= l + sizeof(ogg_packet);
 229         memmove(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
 230 //        av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
 231     }
 232
 233     return l;
 234 }
 235
 236 static av_cold int oggvorbis_encode_close(AVCodecContext *avccontext)
 237 {
 238     OggVorbisContext *context = avccontext->priv_data;
 239 /*  ogg_packet op ; */
 240
 241     vorbis_analysis_wrote(&context->vd, 0);  /* notify vorbisenc this is EOF */
 242
 243     vorbis_block_clear(&context->vb);
 244     vorbis_dsp_clear(&context->vd);
 245     vorbis_info_clear(&context->vi);
 246
 247     av_freep(&avccontext->coded_frame);
 248     av_freep(&avccontext->extradata);
 249
 250     return 0;
 251 }
 252
 253 AVCodec ff_libvorbis_encoder = {
 254     .name           = "libvorbis",
 255     .type           = AVMEDIA_TYPE_AUDIO,
 256     .id             = CODEC_ID_VORBIS,
 257     .priv_data_size = sizeof(OggVorbisContext),
 258     .init           = oggvorbis_encode_init,
 259     .encode         = oggvorbis_encode_frame,
 260     .close          = oggvorbis_encode_close,
 261     .capabilities   = CODEC_CAP_DELAY,
 262     .sample_fmts    = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE },
 263     .long_name      = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
 264     .priv_class     = &class,
 265 };