ffmpeg/libavcodec/libvorbis.c

334 lines
12 KiB
C

/*
* copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
*
* This file is part of FFmpeg.
*
* FFmpeg 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,
* 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
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file
* Ogg Vorbis codec support via libvorbisenc.
* @author Mark Hills <mark@pogo.org.uk>
*/
#include <vorbis/vorbisenc.h>
#include "libavutil/opt.h"
#include "avcodec.h"
#include "bytestream.h"
#include "internal.h"
#include "vorbis.h"
#undef NDEBUG
#include <assert.h>
#define OGGVORBIS_FRAME_SIZE 64
#define BUFFER_SIZE (1024 * 64)
typedef struct OggVorbisContext {
AVClass *av_class;
vorbis_info vi;
vorbis_dsp_state vd;
vorbis_block vb;
uint8_t buffer[BUFFER_SIZE];
int buffer_index;
int eof;
/* decoder */
vorbis_comment vc;
ogg_packet op;
double iblock;
} 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 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 av_cold int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext)
{
OggVorbisContext *context = avccontext->priv_data;
double cfreq;
int ret;
if (avccontext->flags & CODEC_FLAG_QSCALE) {
/* variable bitrate */
float q = avccontext->global_quality / (float)FF_QP2LAMBDA;
if ((ret = vorbis_encode_setup_vbr(vi, avccontext->channels,
avccontext->sample_rate,
q / 10.0)))
goto error;
} else {
int minrate = avccontext->rc_min_rate > 0 ? avccontext->rc_min_rate : -1;
int maxrate = avccontext->rc_min_rate > 0 ? avccontext->rc_max_rate : -1;
/* constant bitrate */
if ((ret = vorbis_encode_setup_managed(vi, avccontext->channels,
avccontext->sample_rate, minrate,
avccontext->bit_rate, maxrate)))
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; /* should not happen */
}
/* cutoff frequency */
if (avccontext->cutoff > 0) {
cfreq = avccontext->cutoff / 1000.0;
if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq)))
goto error; /* should not happen */
}
if (context->iblock) {
if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &context->iblock)))
goto error;
}
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);
}
}
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 *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);
return 0;
}
static av_cold int oggvorbis_encode_init(AVCodecContext *avccontext)
{
OggVorbisContext *context = avccontext->priv_data;
ogg_packet header, header_comm, header_code;
uint8_t *p;
unsigned int offset;
int ret;
vorbis_info_init(&context->vi);
if ((ret = oggvorbis_init_encoder(&context->vi, avccontext))) {
av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed\n");
goto error;
}
if ((ret = vorbis_analysis_init(&context->vd, &context->vi))) {
ret = vorbis_error_to_averror(ret);
goto error;
}
if ((ret = vorbis_block_init(&context->vd, &context->vb))) {
ret = vorbis_error_to_averror(ret);
goto error;
}
vorbis_comment_init(&context->vc);
vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT);
if ((ret = vorbis_analysis_headerout(&context->vd, &context->vc, &header,
&header_comm, &header_code))) {
ret = vorbis_error_to_averror(ret);
goto error;
}
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);
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);
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;
assert(offset == avccontext->extradata_size);
#if 0
vorbis_block_clear(&context->vb);
vorbis_dsp_clear(&context->vd);
vorbis_info_clear(&context->vi);
#endif
vorbis_comment_clear(&context->vc);
avccontext->frame_size = OGGVORBIS_FRAME_SIZE;
avccontext->coded_frame = avcodec_alloc_frame();
if (!avccontext->coded_frame) {
ret = AVERROR(ENOMEM);
goto error;
}
return 0;
error:
oggvorbis_encode_close(avccontext);
return ret;
}
static int oggvorbis_encode_frame(AVCodecContext *avccontext,
unsigned char *packets,
int buf_size, void *data)
{
OggVorbisContext *context = avccontext->priv_data;
ogg_packet op;
signed short *audio = data;
int l;
if (data) {
const int samples = avccontext->frame_size;
float **buffer;
int c, channels = context->vi.channels;
buffer = vorbis_analysis_buffer(&context->vd, samples);
for (c = 0; c < channels; c++) {
int co = (channels > 8) ? c :
ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
for (l = 0; l < samples; l++)
buffer[c][l] = audio[l * channels + co] / 32768.f;
}
vorbis_analysis_wrote(&context->vd, samples);
} else {
if (!context->eof)
vorbis_analysis_wrote(&context->vd, 0);
context->eof = 1;
}
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 && op.e_o_s)
continue;
if (context->buffer_index + sizeof(ogg_packet) + op.bytes > BUFFER_SIZE) {
av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.\n");
return AVERROR(EINVAL);
}
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);
}
}
l = 0;
if (context->buffer_index) {
ogg_packet *op2 = (ogg_packet *)context->buffer;
op2->packet = context->buffer + sizeof(ogg_packet);
l = op2->bytes;
avccontext->coded_frame->pts = ff_samples_to_time_base(avccontext,
op2->granulepos);
//FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
if (l > buf_size) {
av_log(avccontext, AV_LOG_ERROR, "libvorbis: buffer overflow.\n");
return AVERROR(EINVAL);
}
memcpy(packets, op2->packet, l);
context->buffer_index -= l + sizeof(ogg_packet);
memmove(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
// av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
}
return l;
}
AVCodec ff_libvorbis_encoder = {
.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,
};