X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=modules%2Fffmpeg%2Fproducer%2Faudio%2Faudio_decoder.cpp;h=cd5b377926e30c4ac7ccc71865ce1eb78047a8fc;hb=726897adbf881d3b75f171fff24f2b917ba5f05a;hp=3cd99a38f685aac411bf01c348a98c484f15b1f1;hpb=56c7f363b3a444da24bd1c6a1b877891e4f774a4;p=casparcg diff --git a/modules/ffmpeg/producer/audio/audio_decoder.cpp b/modules/ffmpeg/producer/audio/audio_decoder.cpp index 3cd99a38f..cd5b37792 100644 --- a/modules/ffmpeg/producer/audio/audio_decoder.cpp +++ b/modules/ffmpeg/producer/audio/audio_decoder.cpp @@ -1,186 +1,181 @@ -/* -* copyright (c) 2010 Sveriges Television AB -* -* This file is part of CasparCG. -* -* CasparCG is free software: you can redistribute it and/or modify -* it under the terms of the GNU General Public License as published by -* the Free Software Foundation, either version 3 of the License, or -* (at your option) any later version. -* -* CasparCG 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 General Public License for more details. - -* You should have received a copy of the GNU General Public License -* along with CasparCG. If not, see . -* -*/ -#include "../../stdafx.h" - -#include "audio_decoder.h" - -#include - -#if defined(_MSC_VER) -#pragma warning (push) -#pragma warning (disable : 4244) -#endif -extern "C" -{ - #define __STDC_CONSTANT_MACROS - #define __STDC_LIMIT_MACROS - #include - #include -} -#if defined(_MSC_VER) -#pragma warning (pop) -#endif - -namespace caspar { - -struct audio_decoder::implementation : boost::noncopyable -{ - std::shared_ptr codec_context_; - const core::video_format_desc format_desc_; - int index_; - std::shared_ptr resampler_; - - std::vector> buffer1_; - std::vector> buffer2_; - std::vector> audio_samples_; - std::queue> packets_; - - int64_t nb_frames_; -public: - explicit implementation(const std::shared_ptr& context, const core::video_format_desc& format_desc) - : format_desc_(format_desc) - , nb_frames_(0) - { - AVCodec* dec; - index_ = av_find_best_stream(context.get(), AVMEDIA_TYPE_AUDIO, -1, -1, &dec, 0); - - if(index_ < 0) - return; - - int errn = avcodec_open(context->streams[index_]->codec, dec); - if(errn < 0) - return; - - codec_context_.reset(context->streams[index_]->codec, avcodec_close); - - //nb_frames_ = context->streams[index_]->nb_frames; - //if(nb_frames_ == 0) - // nb_frames_ = context->streams[index_]->duration * context->streams[index_]->time_base.den; - - if(codec_context_ && - (codec_context_->sample_rate != static_cast(format_desc_.audio_sample_rate) || - codec_context_->channels != static_cast(format_desc_.audio_channels)) || - codec_context_->sample_fmt != AV_SAMPLE_FMT_S16) - { - auto resampler = av_audio_resample_init(format_desc_.audio_channels, codec_context_->channels, - format_desc_.audio_sample_rate, codec_context_->sample_rate, - AV_SAMPLE_FMT_S16, codec_context_->sample_fmt, - 16, 10, 0, 0.8); - - CASPAR_LOG(warning) << L" Invalid audio format. Resampling."; - - if(resampler) - resampler_.reset(resampler, audio_resample_close); - else - codec_context_ = nullptr; - } - } - - void push(const std::shared_ptr& packet) - { - if(!codec_context_) - return; - - if(packet && packet->stream_index != index_) - return; - - packets_.push(packet); - } - - std::vector>> poll() - { - std::vector>> result; - - if(!codec_context_) - result.push_back(std::make_shared>(format_desc_.audio_samples_per_frame, 0)); - else if(!packets_.empty()) - { - auto packet = std::move(packets_.front()); - packets_.pop(); - - if(packet) - { - AVPacket pkt; - av_init_packet(&pkt); - pkt.data = packet->data; - pkt.size = packet->size; - - for(int n = 0; n < 64 && pkt.size > 0; ++n) - result.push_back(decode(pkt)); - } - else - { - avcodec_flush_buffers(codec_context_.get()); - result.push_back(nullptr); - } - } - - return result; - } - - std::shared_ptr> decode(AVPacket& pkt) - { - buffer1_.resize(AVCODEC_MAX_AUDIO_FRAME_SIZE*2, 0); - int written_bytes = buffer1_.size() - FF_INPUT_BUFFER_PADDING_SIZE; - - const int ret = avcodec_decode_audio3(codec_context_.get(), reinterpret_cast(buffer1_.data()), &written_bytes, &pkt); - if(ret < 0) - { - BOOST_THROW_EXCEPTION( - invalid_operation() << - boost::errinfo_api_function("avcodec_decode_audio2") << - boost::errinfo_errno(AVUNERROR(ret))); - } - - // There might be several frames in one packet. - pkt.size -= ret; - pkt.data += ret; - - buffer1_.resize(written_bytes); - - if(resampler_) - { - buffer2_.resize(AVCODEC_MAX_AUDIO_FRAME_SIZE*2, 0); - auto ret = audio_resample(resampler_.get(), - reinterpret_cast(buffer2_.data()), - reinterpret_cast(buffer1_.data()), - buffer1_.size() / (av_get_bytes_per_sample(codec_context_->sample_fmt) * codec_context_->channels)); - buffer2_.resize(ret * av_get_bytes_per_sample(AV_SAMPLE_FMT_S16) * format_desc_.audio_channels); - std::swap(buffer1_, buffer2_); - } - - const auto n_samples = buffer1_.size() / av_get_bytes_per_sample(AV_SAMPLE_FMT_S16); - const auto samples = reinterpret_cast(buffer1_.data()); - - return std::make_shared>(samples, samples + n_samples); - } - - bool ready() const - { - return !codec_context_ || !packets_.empty(); - } -}; - -audio_decoder::audio_decoder(const std::shared_ptr& context, const core::video_format_desc& format_desc) : impl_(new implementation(context, format_desc)){} -void audio_decoder::push(const std::shared_ptr& packet){impl_->push(packet);} -bool audio_decoder::ready() const{return impl_->ready();} -std::vector>> audio_decoder::poll(){return impl_->poll();} -int64_t audio_decoder::nb_frames() const{return impl_->nb_frames_;} -} \ No newline at end of file +/* +* Copyright 2013 Sveriges Television AB http://casparcg.com/ +* +* This file is part of CasparCG (www.casparcg.com). +* +* CasparCG is free software: you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation, either version 3 of the License, or +* (at your option) any later version. +* +* CasparCG 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 General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with CasparCG. If not, see . +* +* Author: Robert Nagy, ronag89@gmail.com +*/ + +#include "../../stdafx.h" + +#include "audio_decoder.h" + +#include "../util/util.h" +#include "../../ffmpeg_error.h" + +#include +#include + +#include + +#include + +#if defined(_MSC_VER) +#pragma warning (push) +#pragma warning (disable : 4244) +#endif +extern "C" +{ + #include + #include + #include +} +#if defined(_MSC_VER) +#pragma warning (pop) +#endif + +namespace caspar { namespace ffmpeg { + +struct audio_decoder::implementation : boost::noncopyable +{ + int index_ = -1; + const spl::shared_ptr codec_context_; + const int out_samplerate_; + + cache_aligned_vector buffer_; + + std::queue> packets_; + + std::shared_ptr swr_ { + swr_alloc_set_opts( + nullptr, + codec_context_->channel_layout + ? codec_context_->channel_layout + : av_get_default_channel_layout(codec_context_->channels), + AV_SAMPLE_FMT_S32, + out_samplerate_, + codec_context_->channel_layout + ? codec_context_->channel_layout + : av_get_default_channel_layout(codec_context_->channels), + codec_context_->sample_fmt, + codec_context_->sample_rate, + 0, + nullptr), + [](SwrContext* p) + { + swr_free(&p); + } + }; + +public: + explicit implementation(const spl::shared_ptr& context, int out_samplerate) + : codec_context_(open_codec(*context, AVMEDIA_TYPE_AUDIO, index_, false)) + , out_samplerate_(out_samplerate) + , buffer_(10 * out_samplerate_ * codec_context_->channels) // 10 seconds of audio + { + if(!swr_) + BOOST_THROW_EXCEPTION(bad_alloc()); + + THROW_ON_ERROR2(swr_init(swr_.get()), "[audio_decoder]"); + + codec_context_->refcounted_frames = 1; + } + + void push(const std::shared_ptr& packet) + { + if(!packet) + return; + + if(packet->stream_index == index_ || packet->data == nullptr) + packets_.push(spl::make_shared_ptr(packet)); + } + + std::shared_ptr poll() + { + if(packets_.empty()) + return nullptr; + + auto packet = packets_.front(); + + if(packet->data == nullptr) + { + packets_.pop(); + avcodec_flush_buffers(codec_context_.get()); + return flush_audio(); + } + + auto audio = decode(*packet); + + if(packet->size == 0) + packets_.pop(); + + return audio; + } + + std::shared_ptr decode(AVPacket& pkt) + { + auto decoded_frame = create_frame(); + + int got_frame = 0; + auto len = THROW_ON_ERROR2(avcodec_decode_audio4(codec_context_.get(), decoded_frame.get(), &got_frame, &pkt), "[audio_decoder]"); + + if (len == 0) + { + pkt.size = 0; + return nullptr; + } + + pkt.data += len; + pkt.size -= len; + + if (!got_frame) + return nullptr; + + const uint8_t **in = const_cast(decoded_frame->extended_data); + uint8_t* out[] = { reinterpret_cast(buffer_.data()) }; + + const auto channel_samples = swr_convert( + swr_.get(), + out, + static_cast(buffer_.size()) / codec_context_->channels, + in, + decoded_frame->nb_samples); + + return std::make_shared( + buffer_.begin(), + buffer_.begin() + channel_samples * decoded_frame->channels); + } + + bool ready() const + { + return packets_.size() > 10; + } + + std::wstring print() const + { + return L"[audio-decoder] " + u16(codec_context_->codec->long_name); + } +}; + +audio_decoder::audio_decoder(const spl::shared_ptr& context, int out_samplerate) : impl_(new implementation(context, out_samplerate)){} +void audio_decoder::push(const std::shared_ptr& packet){impl_->push(packet);} +bool audio_decoder::ready() const{return impl_->ready();} +std::shared_ptr audio_decoder::poll() { return impl_->poll(); } +int audio_decoder::num_channels() const { return impl_->codec_context_->channels; } +uint64_t audio_decoder::ffmpeg_channel_layout() const { return impl_->codec_context_->channel_layout; } +std::wstring audio_decoder::print() const{return impl_->print();} + +}}