int64_t swr_get_delay(struct SwrContext *s, int64_t base)
{
int64_t in_sr, out_sr;
av_opt_get_int(s, "in_sample_rate", 0, &in_sr);
av_opt_get_int(s, "out_sample_rate", 0, &out_sr);
return av_rescale_rnd(avresample_available(s), base, out_sr, AV_ROUND_UP) + av_rescale_rnd(avresample_get_delay(s), base, in_sr, AV_ROUND_UP);
}
/**
* Allocates and initializes the encoder context and frame in FFAudioIO.
* As parameters serve the output parameters of the SwrContext from FFAudioIO.
* Therefore the SwrContext must be setup first for this to be successful.
*
* @param env JNIEnv
* @param aio FFAudioIO (our context)
* @param encoder AVCodec to use to setup the encoder AVCodecContext
* @return a negative value, if something goes wrong
*/
int ff_init_encoder(JNIEnv *env, FFAudioIO *aio, AVCodec *encoder) {
int res = 0;
int64_t out_sample_rate;
int64_t out_channel_count;
int64_t out_channel_layout;
enum AVSampleFormat out_sample_fmt;
// make sure we clean up before resetting this
// in case this is called twice
if (aio->encode_frame) {
av_frame_free(&aio->encode_frame);
}
if (aio->encode_context) {
avcodec_close(aio->encode_context);
av_free(aio->encode_context);
}
aio->encode_context = avcodec_alloc_context3(encoder);
if (!aio->encode_context) {
res = AVERROR(ENOMEM);
throwIOExceptionIfError(env, res, "Could not allocate codec context.");
goto bail;
}
// init to whatever we have in SwrContext
av_opt_get_int(aio->swr_context, "out_channel_count", 0, &out_channel_count);
av_opt_get_int(aio->swr_context, "out_channel_layout", 0, &out_channel_layout);
av_opt_get_int(aio->swr_context, "out_sample_rate", 0, &out_sample_rate);
av_opt_get_sample_fmt(aio->swr_context, "out_sample_fmt", 0, &out_sample_fmt);
aio->encode_context->sample_fmt = out_sample_fmt;
aio->encode_context->sample_rate = out_sample_rate;
aio->encode_context->channel_layout = out_channel_layout;
aio->encode_context->channels = out_channel_count;
res = avcodec_open2(aio->encode_context, encoder, NULL);
if (res < 0) {
res = AVERROR(ENOMEM);
throwIOExceptionIfError(env, res, "Could not open encoder.");
goto bail;
}
aio->encode_frame = av_frame_alloc();
if (!aio->encode_frame) {
res = AVERROR(ENOMEM);
throwIOExceptionIfError(env, res, "Could not allocate encoder frame.");
goto bail;
}
aio->encode_frame->nb_samples = aio->encode_context->frame_size; // this will be changed later!!
aio->encode_frame->format = aio->encode_context->sample_fmt;
aio->encode_frame->channel_layout = aio->encode_context->channel_layout;
bail:
return res;
}
int ff_rtp_get_payload_type(AVFormatContext *fmt, AVCodecContext *codec)
{
int i;
AVOutputFormat *ofmt = fmt ? fmt->oformat : NULL;
/* Was the payload type already specified for the RTP muxer? */
if (ofmt && ofmt->priv_class) {
int64_t payload_type;
if (av_opt_get_int(fmt->priv_data, "payload_type", 0, &payload_type) >= 0 &&
payload_type >= 0)
return (int)payload_type;
}
/* static payload type */
for (i = 0; AVRtpPayloadTypes[i].pt >= 0; ++i)
if (AVRtpPayloadTypes[i].codec_id == codec->codec_id) {
if (codec->codec_id == CODEC_ID_H263 && (!fmt ||
!fmt->oformat->priv_class ||
!av_opt_flag_is_set(fmt->priv_data, "rtpflags", "rfc2190")))
continue;
if (codec->codec_id == CODEC_ID_PCM_S16BE)
if (codec->channels != AVRtpPayloadTypes[i].audio_channels)
continue;
return AVRtpPayloadTypes[i].pt;
}
/* dynamic payload type */
return RTP_PT_PRIVATE + (codec->codec_type == AVMEDIA_TYPE_AUDIO);
}
/**
* Resample a buffer using FFAudioUI->swr_context.
* The returned out buffer needs to be freed by the caller.
*
* @param aio FFAudioIO context
* @param out_buf out buffer
* @param out_samples out samples
* @param in_buf in buffer
* @param in_samples in samples
* @return number of samples copied/converted or a negative value, should things go wrong
*/
static int resample(FFAudioIO *aio, uint8_t **out_buf, int out_samples, const uint8_t **in_buf, const int in_samples) {
int res = 0;
int64_t out_channel_count;
enum AVSampleFormat out_sample_format;
if (out_samples == 0) goto bail; // nothing to do.
av_opt_get_int(aio->swr_context, "out_channel_count", 0, &out_channel_count);
av_opt_get_sample_fmt(aio->swr_context, "out_sample_fmt", 0, &out_sample_format);
#ifdef DEBUG
fprintf(stderr, "resample: out_samples=%d in_samples=%d, channels=%d sample_format=%d\n",
out_samples, in_samples, (int)out_channel_count, out_sample_format);
#endif
// allocate temp buffer for resampled data
res = av_samples_alloc(out_buf, NULL, out_channel_count, out_samples, out_sample_format, 1);
if (res < 0) {
res = AVERROR(ENOMEM);
throwIOExceptionIfError(aio->env, res, "Could not allocate resample buffer.");
goto bail;
}
// run the SWR conversion (even if it is not strictly necessary)
res = swr_convert(aio->swr_context, out_buf, out_samples, in_buf, in_samples);
if (res < 0) {
throwIOExceptionIfError(aio->env, res, "Failed to convert audio data.");
goto bail;
}
bail:
return res;
}
int Option::getInt() const
{
int64_t out_val;
int error = av_opt_get_int( _avContext, getName().c_str(), AV_OPT_SEARCH_CHILDREN, &out_val );
checkFFmpegGetOption( error );
return out_val;
}
void Option::setFlag( const std::string& flag, const bool enable )
{
int64_t optVal;
int error = av_opt_get_int( _avContext, getName().c_str(), AV_OPT_SEARCH_CHILDREN, &optVal );
checkFFmpegGetOption( error );
if( enable )
optVal = optVal | _avOption->default_val.i64;
else
optVal = optVal &~ _avOption->default_val.i64;
error = av_opt_set_int( _avContext, getName().c_str(), optVal, AV_OPT_SEARCH_CHILDREN );
checkFFmpegSetOption( error, flag );
}
void setSampleRate(int sr)
{
if (swr)
{
int64_t outSampleRate;
av_opt_get_int(swr, "out_sample_rate", 0, &outSampleRate);
if (outSampleRate != host.sampleRate)
{
av_opt_set_int(swr, "out_sample_rate", host.sampleRate, 0);
swr_init(swr);
JIF(swr_init(swr), "failed to init audio resampler.");
}
}
host.sampleRate = sr;
return;
err:
host.state = CodecBoxDecoderState::Failed;
close();
}
int ff_rtp_get_payload_type(AVFormatContext *fmt,
AVCodecContext *codec, int idx)
{
int i;
AVOutputFormat *ofmt = fmt ? fmt->oformat : NULL;
/* Was the payload type already specified for the RTP muxer? */
if (ofmt && ofmt->priv_class && fmt->priv_data) {
int64_t payload_type;
if (av_opt_get_int(fmt->priv_data, "payload_type", 0, &payload_type) >= 0 &&
payload_type >= 0)
return (int)payload_type;
}
/* static payload type */
for (i = 0; rtp_payload_types[i].pt >= 0; ++i)
if (rtp_payload_types[i].codec_id == codec->codec_id) {
if (codec->codec_id == AV_CODEC_ID_H263 && (!fmt || !fmt->oformat ||
!fmt->oformat->priv_class || !fmt->priv_data ||
!av_opt_flag_is_set(fmt->priv_data, "rtpflags", "rfc2190")))
continue;
/* G722 has 8000 as nominal rate even if the sample rate is 16000,
* see section 4.5.2 in RFC 3551. */
if (codec->codec_id == AV_CODEC_ID_ADPCM_G722 &&
codec->sample_rate == 16000 && codec->channels == 1)
return rtp_payload_types[i].pt;
if (codec->codec_type == AVMEDIA_TYPE_AUDIO &&
((rtp_payload_types[i].clock_rate > 0 &&
codec->sample_rate != rtp_payload_types[i].clock_rate) ||
(rtp_payload_types[i].audio_channels > 0 &&
codec->channels != rtp_payload_types[i].audio_channels)))
continue;
return rtp_payload_types[i].pt;
}
if (idx < 0)
idx = codec->codec_type == AVMEDIA_TYPE_AUDIO;
/* dynamic payload type */
return RTP_PT_PRIVATE + idx;
}
static int opus_decode_subpacket(OpusStreamContext *s,
const uint8_t *buf, int buf_size,
int nb_samples)
{
int output_samples = 0;
int flush_needed = 0;
int i, j, ret;
/* check if we need to flush the resampler */
if (swr_is_initialized(s->swr)) {
if (buf) {
int64_t cur_samplerate;
av_opt_get_int(s->swr, "in_sample_rate", 0, &cur_samplerate);
flush_needed = (s->packet.mode == OPUS_MODE_CELT) || (cur_samplerate != s->silk_samplerate);
} else {
flush_needed = !!s->delayed_samples;
}
}
if (!buf && !flush_needed)
return 0;
/* use dummy output buffers if the channel is not mapped to anything */
if (!s->out[0] ||
(s->output_channels == 2 && !s->out[1])) {
av_fast_malloc(&s->out_dummy, &s->out_dummy_allocated_size, s->out_size);
if (!s->out_dummy)
return AVERROR(ENOMEM);
if (!s->out[0])
s->out[0] = s->out_dummy;
if (!s->out[1])
s->out[1] = s->out_dummy;
}
/* flush the resampler if necessary */
if (flush_needed) {
ret = opus_flush_resample(s, s->delayed_samples);
if (ret < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Error flushing the resampler.\n");
return ret;
}
swr_close(s->swr);
output_samples += s->delayed_samples;
s->delayed_samples = 0;
if (!buf)
goto finish;
}
/* decode all the frames in the packet */
for (i = 0; i < s->packet.frame_count; i++) {
int size = s->packet.frame_size[i];
int samples = opus_decode_frame(s, buf + s->packet.frame_offset[i], size);
if (samples < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Error decoding an Opus frame.\n");
if (s->avctx->err_recognition & AV_EF_EXPLODE)
return samples;
for (j = 0; j < s->output_channels; j++)
memset(s->out[j], 0, s->packet.frame_duration * sizeof(float));
samples = s->packet.frame_duration;
}
output_samples += samples;
for (j = 0; j < s->output_channels; j++)
s->out[j] += samples;
s->out_size -= samples * sizeof(float);
}
finish:
s->out[0] = s->out[1] = NULL;
s->out_size = 0;
return output_samples;
}
/**
* Decode a frame to a packet, run the result through SwrContext, if desired, encode it via an appropriate
* encoder, and write the results to the Java-side native buffer.
*
* @param aio FFAudio context
* @param cached true or false
* @return number of bytes placed into java buffer or a negative value, if something went wrong
*/
static int decode_packet(FFAudioIO *aio, int cached) {
int res = 0;
uint8_t **resample_buf = NULL;
jobject byte_buffer = NULL;
uint8_t *javaBuffer = NULL;
uint32_t out_buf_size = 0;
int out_buf_samples = 0;
int64_t out_channel_count;
int64_t out_sample_rate;
int flush = aio->got_frame;
enum AVSampleFormat out;
int bytesConsumed = 0;
init_ids(aio->env, aio->java_instance);
av_opt_get_int(aio->swr_context, "out_channel_count", 0, &out_channel_count);
av_opt_get_int(aio->swr_context, "out_sample_rate", 0, &out_sample_rate);
av_opt_get_sample_fmt(aio->swr_context, "out_sample_fmt", 0, &out);
resample_buf = av_mallocz(sizeof(uint8_t *) * 1); // one plane!
// make sure we really have an audio packet
if (aio->decode_packet.stream_index == aio->stream_index) {
// decode frame
// got_frame indicates whether we got a frame
bytesConsumed = avcodec_decode_audio4(aio->decode_context, aio->decode_frame, &aio->got_frame, &aio->decode_packet);
if (bytesConsumed < 0) {
throwUnsupportedAudioFileExceptionIfError(aio->env, bytesConsumed, "Failed to decode audio frame.");
return bytesConsumed;
}
if (aio->got_frame) {
aio->decoded_samples += aio->decode_frame->nb_samples;
out_buf_samples = aio->decode_frame->nb_samples;
#ifdef DEBUG
fprintf(stderr, "samples%s n:%" PRIu64 " nb_samples:%d pts:%s\n",
cached ? "(cached)" : "",
aio->decoded_samples, aio->decode_frame->nb_samples,
av_ts2timestr(aio->decode_frame->pts, &aio->decode_context->time_base));
#endif
// adjust out sample number for a different sample rate
// this is an estimate!!
out_buf_samples = av_rescale_rnd(
swr_get_delay(aio->swr_context, aio->stream->codecpar->sample_rate) + aio->decode_frame->nb_samples,
out_sample_rate,
aio->stream->codecpar->sample_rate,
AV_ROUND_UP
);
// allocate new aio->audio_data buffers
res = av_samples_alloc(aio->audio_data, NULL, av_frame_get_channels(aio->decode_frame),
aio->decode_frame->nb_samples, aio->decode_frame->format, 1);
if (res < 0) {
throwIOExceptionIfError(aio->env, res, "Could not allocate audio buffer.");
return AVERROR(ENOMEM);
}
// copy audio data to aio->audio_data
av_samples_copy(aio->audio_data, aio->decode_frame->data, 0, 0,
aio->decode_frame->nb_samples, av_frame_get_channels(aio->decode_frame), aio->decode_frame->format);
res = resample(aio, resample_buf, out_buf_samples, (const uint8_t **)aio->audio_data, aio->decode_frame->nb_samples);
if (res < 0) goto bail;
else out_buf_samples = res;
} else if (flush && swr_get_delay(aio->swr_context, aio->stream->codecpar->sample_rate)) {
res = resample(aio, resample_buf, swr_get_delay(aio->swr_context, aio->stream->codecpar->sample_rate), NULL, 0);
if (res < 0) goto bail;
else out_buf_samples = res;
} else {
#ifdef DEBUG
fprintf(stderr, "Got no frame.\n");
#endif
}
if (out_buf_samples > 0) {
res = av_samples_get_buffer_size(NULL, (int)out_channel_count, out_buf_samples, out, 1);
if (res < 0) goto bail;
else out_buf_size = res;
// ensure native buffer capacity
if (aio->java_buffer_capacity < out_buf_size) {
aio->java_buffer_capacity = (*aio->env)->CallIntMethod(aio->env, aio->java_instance, setNativeBufferCapacity_MID, (jint)out_buf_size);
}
// get java-managed byte buffer reference
byte_buffer = (*aio->env)->GetObjectField(aio->env, aio->java_instance, nativeBuffer_FID);
if (!byte_buffer) {
res = -1;
throwIOExceptionIfError(aio->env, 1, "Failed to get native buffer.");
goto bail;
}
// we have some samples, let's copy them to the java buffer, using the desired encoding
javaBuffer = (uint8_t *)(*aio->env)->GetDirectBufferAddress(aio->env, byte_buffer);
if (!javaBuffer) {
throwIOExceptionIfError(aio->env, 1, "Failed to get address for native buffer.");
goto bail;
}
if (aio->encode_context) {
aio->encode_frame->nb_samples = out_buf_samples;
res = encode_buffer(aio, resample_buf[0], out_buf_size, javaBuffer);
if (res < 0) {
out_buf_size = 0;
goto bail;
}
out_buf_size = res;
} else {
memcpy(javaBuffer, resample_buf[0], out_buf_size);
}
// we already wrote to the buffer, now we still need to
// set new bytebuffer limit and position to 0.
(*aio->env)->CallObjectMethod(aio->env, byte_buffer, rewind_MID);
(*aio->env)->CallObjectMethod(aio->env, byte_buffer, limit_MID, out_buf_size);
}
}
aio->resampled_samples += out_buf_size;
bail:
if (resample_buf) {
if (resample_buf[0]) av_freep(&resample_buf[0]);
av_free(resample_buf);
}
if (aio->audio_data[0]) av_freep(&aio->audio_data[0]);
return res;
}
static int opus_decode_frame(OpusStreamContext *s, const uint8_t *data, int size)
{
int samples = s->packet.frame_duration;
int redundancy = 0;
int redundancy_size, redundancy_pos;
int ret, i, consumed;
int delayed_samples = s->delayed_samples;
ret = ff_opus_rc_dec_init(&s->rc, data, size);
if (ret < 0)
return ret;
/* decode the silk frame */
if (s->packet.mode == OPUS_MODE_SILK || s->packet.mode == OPUS_MODE_HYBRID) {
#if CONFIG_SWRESAMPLE
if (!swr_is_initialized(s->swr)) {
#elif CONFIG_AVRESAMPLE
if (!avresample_is_open(s->avr)) {
#endif
ret = opus_init_resample(s);
if (ret < 0)
return ret;
}
samples = ff_silk_decode_superframe(s->silk, &s->rc, s->silk_output,
FFMIN(s->packet.bandwidth, OPUS_BANDWIDTH_WIDEBAND),
s->packet.stereo + 1,
silk_frame_duration_ms[s->packet.config]);
if (samples < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Error decoding a SILK frame.\n");
return samples;
}
#if CONFIG_SWRESAMPLE
samples = swr_convert(s->swr,
(uint8_t**)s->out, s->packet.frame_duration,
(const uint8_t**)s->silk_output, samples);
#elif CONFIG_AVRESAMPLE
samples = avresample_convert(s->avr, (uint8_t**)s->out, s->out_size,
s->packet.frame_duration,
(uint8_t**)s->silk_output,
sizeof(s->silk_buf[0]),
samples);
#endif
if (samples < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Error resampling SILK data.\n");
return samples;
}
av_assert2((samples & 7) == 0);
s->delayed_samples += s->packet.frame_duration - samples;
} else
ff_silk_flush(s->silk);
// decode redundancy information
consumed = opus_rc_tell(&s->rc);
if (s->packet.mode == OPUS_MODE_HYBRID && consumed + 37 <= size * 8)
redundancy = ff_opus_rc_dec_log(&s->rc, 12);
else if (s->packet.mode == OPUS_MODE_SILK && consumed + 17 <= size * 8)
redundancy = 1;
if (redundancy) {
redundancy_pos = ff_opus_rc_dec_log(&s->rc, 1);
if (s->packet.mode == OPUS_MODE_HYBRID)
redundancy_size = ff_opus_rc_dec_uint(&s->rc, 256) + 2;
else
redundancy_size = size - (consumed + 7) / 8;
size -= redundancy_size;
if (size < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Invalid redundancy frame size.\n");
return AVERROR_INVALIDDATA;
}
if (redundancy_pos) {
ret = opus_decode_redundancy(s, data + size, redundancy_size);
if (ret < 0)
return ret;
ff_celt_flush(s->celt);
}
}
/* decode the CELT frame */
if (s->packet.mode == OPUS_MODE_CELT || s->packet.mode == OPUS_MODE_HYBRID) {
float *out_tmp[2] = { s->out[0], s->out[1] };
float **dst = (s->packet.mode == OPUS_MODE_CELT) ?
out_tmp : s->celt_output;
int celt_output_samples = samples;
int delay_samples = av_audio_fifo_size(s->celt_delay);
if (delay_samples) {
if (s->packet.mode == OPUS_MODE_HYBRID) {
av_audio_fifo_read(s->celt_delay, (void**)s->celt_output, delay_samples);
for (i = 0; i < s->output_channels; i++) {
s->fdsp->vector_fmac_scalar(out_tmp[i], s->celt_output[i], 1.0,
delay_samples);
out_tmp[i] += delay_samples;
}
celt_output_samples -= delay_samples;
} else {
av_log(s->avctx, AV_LOG_WARNING,
"Spurious CELT delay samples present.\n");
av_audio_fifo_drain(s->celt_delay, delay_samples);
if (s->avctx->err_recognition & AV_EF_EXPLODE)
return AVERROR_BUG;
}
}
ff_opus_rc_dec_raw_init(&s->rc, data + size, size);
ret = ff_celt_decode_frame(s->celt, &s->rc, dst,
s->packet.stereo + 1,
s->packet.frame_duration,
(s->packet.mode == OPUS_MODE_HYBRID) ? 17 : 0,
ff_celt_band_end[s->packet.bandwidth]);
if (ret < 0)
return ret;
if (s->packet.mode == OPUS_MODE_HYBRID) {
int celt_delay = s->packet.frame_duration - celt_output_samples;
void *delaybuf[2] = { s->celt_output[0] + celt_output_samples,
s->celt_output[1] + celt_output_samples };
for (i = 0; i < s->output_channels; i++) {
s->fdsp->vector_fmac_scalar(out_tmp[i],
s->celt_output[i], 1.0,
celt_output_samples);
}
ret = av_audio_fifo_write(s->celt_delay, delaybuf, celt_delay);
if (ret < 0)
return ret;
}
} else
ff_celt_flush(s->celt);
if (s->redundancy_idx) {
for (i = 0; i < s->output_channels; i++)
opus_fade(s->out[i], s->out[i],
s->redundancy_output[i] + 120 + s->redundancy_idx,
ff_celt_window2 + s->redundancy_idx, 120 - s->redundancy_idx);
s->redundancy_idx = 0;
}
if (redundancy) {
if (!redundancy_pos) {
ff_celt_flush(s->celt);
ret = opus_decode_redundancy(s, data + size, redundancy_size);
if (ret < 0)
return ret;
for (i = 0; i < s->output_channels; i++) {
opus_fade(s->out[i] + samples - 120 + delayed_samples,
s->out[i] + samples - 120 + delayed_samples,
s->redundancy_output[i] + 120,
ff_celt_window2, 120 - delayed_samples);
if (delayed_samples)
s->redundancy_idx = 120 - delayed_samples;
}
} else {
for (i = 0; i < s->output_channels; i++) {
memcpy(s->out[i] + delayed_samples, s->redundancy_output[i], 120 * sizeof(float));
opus_fade(s->out[i] + 120 + delayed_samples,
s->redundancy_output[i] + 120,
s->out[i] + 120 + delayed_samples,
ff_celt_window2, 120);
}
}
}
return samples;
}
static int opus_decode_subpacket(OpusStreamContext *s,
const uint8_t *buf, int buf_size,
float **out, int out_size,
int nb_samples)
{
int output_samples = 0;
int flush_needed = 0;
int i, j, ret;
s->out[0] = out[0];
s->out[1] = out[1];
s->out_size = out_size;
/* check if we need to flush the resampler */
#if CONFIG_SWRESAMPLE
if (swr_is_initialized(s->swr)) {
if (buf) {
int64_t cur_samplerate;
av_opt_get_int(s->swr, "in_sample_rate", 0, &cur_samplerate);
flush_needed = (s->packet.mode == OPUS_MODE_CELT) || (cur_samplerate != s->silk_samplerate);
} else {
flush_needed = !!s->delayed_samples;
}
}
#elif CONFIG_AVRESAMPLE
if (avresample_is_open(s->avr)) {
if (buf) {
int64_t cur_samplerate;
av_opt_get_int(s->avr, "in_sample_rate", 0, &cur_samplerate);
flush_needed = (s->packet.mode == OPUS_MODE_CELT) || (cur_samplerate != s->silk_samplerate);
} else {
flush_needed = !!s->delayed_samples;
}
}
#endif
if (!buf && !flush_needed)
return 0;
/* use dummy output buffers if the channel is not mapped to anything */
if (!s->out[0] ||
(s->output_channels == 2 && !s->out[1])) {
av_fast_malloc(&s->out_dummy, &s->out_dummy_allocated_size, s->out_size);
if (!s->out_dummy)
return AVERROR(ENOMEM);
if (!s->out[0])
s->out[0] = s->out_dummy;
if (!s->out[1])
s->out[1] = s->out_dummy;
}
/* flush the resampler if necessary */
if (flush_needed) {
ret = opus_flush_resample(s, s->delayed_samples);
if (ret < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Error flushing the resampler.\n");
return ret;
}
#if CONFIG_SWRESAMPLE
swr_close(s->swr);
#elif CONFIG_AVRESAMPLE
avresample_close(s->avr);
#endif
output_samples += s->delayed_samples;
s->delayed_samples = 0;
if (!buf)
goto finish;
}
/* decode all the frames in the packet */
for (i = 0; i < s->packet.frame_count; i++) {
int size = s->packet.frame_size[i];
int samples = opus_decode_frame(s, buf + s->packet.frame_offset[i], size);
if (samples < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Error decoding an Opus frame.\n");
if (s->avctx->err_recognition & AV_EF_EXPLODE)
return samples;
for (j = 0; j < s->output_channels; j++)
memset(s->out[j], 0, s->packet.frame_duration * sizeof(float));
samples = s->packet.frame_duration;
}
output_samples += samples;
for (j = 0; j < s->output_channels; j++)
s->out[j] += samples;
s->out_size -= samples * sizeof(float);
}
finish:
s->out[0] = s->out[1] = NULL;
s->out_size = 0;
return output_samples;
}
bool D2V::printSettings() {
int stream_type = getStreamType(f->fctx->iformat->name);
int video_id = video_stream->id;
int audio_id = 0;
int64_t ts_packetsize = 0;
if (stream_type == TRANSPORT_STREAM) {
const AVStream *audio_stream = nullptr;
for (unsigned i = 0; i < f->fctx->nb_streams; i++) {
if (f->fctx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
audio_stream = f->fctx->streams[i];
break;
}
}
if (audio_stream)
audio_id = audio_stream->id;
if (av_opt_get_int(f->fctx, "ts_packetsize", AV_OPT_SEARCH_CHILDREN, &ts_packetsize) < 0)
ts_packetsize = 0;
}
int mpeg_type = 0;
if (video_stream->codec->codec_id == AV_CODEC_ID_MPEG1VIDEO)
mpeg_type = 1;
else if (video_stream->codec->codec_id == AV_CODEC_ID_MPEG2VIDEO)
mpeg_type = 2;
int yuvrgb_scale = input_range == ColourRangeLimited ? 1 : 0;
int width, height;
if (av_opt_get_image_size(video_stream->codec, "video_size", 0, &width, &height) < 0)
width = height = -1;
AVRational sar;
if (av_opt_get_q(video_stream->codec, "aspect", 0, &sar) < 0)
sar = { 1, 1 };
AVRational dar = av_mul_q(av_make_q(width, height), sar);
av_reduce(&dar.num, &dar.den, dar.num, dar.den, 1024);
// No AVOption for framerate?
AVRational frame_rate = video_stream->codec->framerate;
std::string settings;
settings += "Stream_Type=" + std::to_string(stream_type) + "\n";
if (stream_type == TRANSPORT_STREAM) {
char pids[100] = { 0 };
snprintf(pids, 100, "%x,%x,%x", video_id, audio_id, 0);
settings += "MPEG2_Transport_PID=";
settings += pids;
settings += "\n";
settings += "Transport_Packet_Size=" + std::to_string(ts_packetsize) + "\n";
}
settings += "MPEG_Type=" + std::to_string(mpeg_type) + "\n";
settings += "iDCT_Algorithm=6\n"; // "32-bit SSEMMX (Skal)". No one cares anyway.
settings += "YUVRGB_Scale=" + std::to_string(yuvrgb_scale) + "\n";
settings += "Luminance_Filter=0,0\n"; // We don't care.
settings += "Clipping=0,0,0,0\n"; // We don't crop here.
settings += "Aspect_Ratio=" + std::to_string(dar.num) + ":" + std::to_string(dar.den) + "\n";
settings += "Picture_Size=" + std::to_string(width) + "x" + std::to_string(height) + "\n";
settings += "Field_Operation=0\n"; // Always tell them honor the pulldown flags.
settings += "Frame_Rate=" + std::to_string((int)((float)frame_rate.num * 1000 / frame_rate.den)) + " (" + std::to_string(frame_rate.num) + "/" + std::to_string(frame_rate.den) + ")\n";
settings += "Location=0,0,0,0\n"; // Whatever.
if (fprintf(d2v_file, "%s", settings.c_str()) < 0) {
error = "Failed to print d2v settings section: fprintf() failed.";
return false;
}
return true;
}
static int config_output(AVFilterLink *outlink)
{
AVFilterContext *ctx = outlink->src;
AVFilterLink *inlink = ctx->inputs[0];
ResampleContext *s = ctx->priv;
char buf1[64], buf2[64];
int ret;
int64_t resampling_forced;
if (s->avr) {
avresample_close(s->avr);
avresample_free(&s->avr);
}
if (inlink->channel_layout == outlink->channel_layout &&
inlink->sample_rate == outlink->sample_rate &&
(inlink->format == outlink->format ||
(av_get_channel_layout_nb_channels(inlink->channel_layout) == 1 &&
av_get_channel_layout_nb_channels(outlink->channel_layout) == 1 &&
av_get_planar_sample_fmt(inlink->format) ==
av_get_planar_sample_fmt(outlink->format))))
return 0;
if (!(s->avr = avresample_alloc_context()))
return AVERROR(ENOMEM);
if (s->options) {
int ret;
AVDictionaryEntry *e = NULL;
while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX)))
av_log(ctx, AV_LOG_VERBOSE, "lavr option: %s=%s\n", e->key, e->value);
ret = av_opt_set_dict(s->avr, &s->options);
if (ret < 0)
return ret;
}
av_opt_set_int(s->avr, "in_channel_layout", inlink ->channel_layout, 0);
av_opt_set_int(s->avr, "out_channel_layout", outlink->channel_layout, 0);
av_opt_set_int(s->avr, "in_sample_fmt", inlink ->format, 0);
av_opt_set_int(s->avr, "out_sample_fmt", outlink->format, 0);
av_opt_set_int(s->avr, "in_sample_rate", inlink ->sample_rate, 0);
av_opt_set_int(s->avr, "out_sample_rate", outlink->sample_rate, 0);
if ((ret = avresample_open(s->avr)) < 0)
return ret;
av_opt_get_int(s->avr, "force_resampling", 0, &resampling_forced);
s->resampling = resampling_forced || (inlink->sample_rate != outlink->sample_rate);
if (s->resampling) {
outlink->time_base = (AVRational){ 1, outlink->sample_rate };
s->next_pts = AV_NOPTS_VALUE;
s->next_in_pts = AV_NOPTS_VALUE;
} else
outlink->time_base = inlink->time_base;
av_get_channel_layout_string(buf1, sizeof(buf1),
-1, inlink ->channel_layout);
av_get_channel_layout_string(buf2, sizeof(buf2),
-1, outlink->channel_layout);
av_log(ctx, AV_LOG_VERBOSE,
"fmt:%s srate:%d cl:%s -> fmt:%s srate:%d cl:%s\n",
av_get_sample_fmt_name(inlink ->format), inlink ->sample_rate, buf1,
av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf2);
return 0;
}
