| /* |
| * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. |
| * |
| * Use of this source code is governed by a BSD-style license |
| * that can be found in the LICENSE file in the root of the source |
| * tree. An additional intellectual property rights grant can be found |
| * in the file PATENTS. All contributing project authors may |
| * be found in the AUTHORS file in the root of the source tree. |
| */ |
| |
| #include "webrtc/modules/audio_coding/neteq4/audio_multi_vector.h" |
| |
| #include <assert.h> |
| |
| #include <algorithm> |
| |
| #include "webrtc/typedefs.h" |
| |
| namespace webrtc { |
| |
| template<typename T> |
| AudioMultiVector<T>::AudioMultiVector(size_t N) { |
| assert(N > 0); |
| if (N < 1) N = 1; |
| for (size_t n = 0; n < N; ++n) { |
| channels_.push_back(new AudioVector<T>); |
| } |
| } |
| |
| template<typename T> |
| AudioMultiVector<T>::AudioMultiVector(size_t N, size_t initial_size) { |
| assert(N > 0); |
| if (N < 1) N = 1; |
| for (size_t n = 0; n < N; ++n) { |
| channels_.push_back(new AudioVector<T>(initial_size)); |
| } |
| } |
| |
| template<typename T> |
| AudioMultiVector<T>::~AudioMultiVector() { |
| typename std::vector<AudioVector<T>*>::iterator it = channels_.begin(); |
| while (it != channels_.end()) { |
| delete (*it); |
| ++it; |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::Clear() { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->Clear(); |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::Zeros(size_t length) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->Clear(); |
| channels_[i]->Extend(length); |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::CopyFrom(AudioMultiVector<T>* copy_to) const { |
| if (copy_to) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->CopyFrom(&(*copy_to)[i]); |
| } |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::PushBackInterleaved(const T* append_this, |
| size_t length) { |
| assert(length % Channels() == 0); |
| size_t length_per_channel = length / Channels(); |
| T* temp_array = new T[length_per_channel]; // Intermediate storage. |
| for (size_t channel = 0; channel < Channels(); ++channel) { |
| // Copy elements to |temp_array|. |
| // Set |source_ptr| to first element of this channel. |
| const T* source_ptr = &append_this[channel]; |
| for (size_t i = 0; i < length_per_channel; ++i) { |
| temp_array[i] = *source_ptr; |
| source_ptr += Channels(); // Jump to next element of this channel. |
| } |
| channels_[channel]->PushBack(temp_array, length_per_channel); |
| } |
| delete [] temp_array; |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::PushBack(const AudioMultiVector<T>& append_this) { |
| assert(Channels() == append_this.Channels()); |
| if (Channels() == append_this.Channels()) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->PushBack(append_this[i]); |
| } |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::PushBackFromIndex( |
| const AudioMultiVector<T>& append_this, |
| size_t index) { |
| assert(index < append_this.Size()); |
| index = std::min(index, append_this.Size() - 1); |
| size_t length = append_this.Size() - index; |
| assert(Channels() == append_this.Channels()); |
| if (Channels() == append_this.Channels()) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->PushBack(&append_this[i][index], length); |
| } |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::PopFront(size_t length) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->PopFront(length); |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::PopBack(size_t length) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->PopBack(length); |
| } |
| } |
| |
| template<typename T> |
| size_t AudioMultiVector<T>::ReadInterleaved(size_t length, |
| T* destination) const { |
| return ReadInterleavedFromIndex(0, length, destination); |
| } |
| |
| template<typename T> |
| size_t AudioMultiVector<T>::ReadInterleavedFromIndex(size_t start_index, |
| size_t length, |
| T* destination) const { |
| if (!destination) { |
| return 0; |
| } |
| size_t index = 0; // Number of elements written to |destination| so far. |
| assert(start_index <= Size()); |
| start_index = std::min(start_index, Size()); |
| if (length + start_index > Size()) { |
| length = Size() - start_index; |
| } |
| for (size_t i = 0; i < length; ++i) { |
| for (size_t channel = 0; channel < Channels(); ++channel) { |
| destination[index] = (*this)[channel][i + start_index]; |
| ++index; |
| } |
| } |
| return index; |
| } |
| |
| template<typename T> |
| size_t AudioMultiVector<T>::ReadInterleavedFromEnd(size_t length, |
| T* destination) const { |
| length = std::min(length, Size()); // Cannot read more than Size() elements. |
| return ReadInterleavedFromIndex(Size() - length, length, destination); |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::OverwriteAt(const AudioMultiVector<T>& insert_this, |
| size_t length, |
| size_t position) { |
| assert(Channels() == insert_this.Channels()); |
| // Cap |length| at the length of |insert_this|. |
| assert(length <= insert_this.Size()); |
| length = std::min(length, insert_this.Size()); |
| if (Channels() == insert_this.Channels()) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->OverwriteAt(&insert_this[i][0], length, position); |
| } |
| } |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::CrossFade(const AudioMultiVector<T>& append_this, |
| size_t fade_length) { |
| assert(Channels() == append_this.Channels()); |
| if (Channels() == append_this.Channels()) { |
| for (size_t i = 0; i < Channels(); ++i) { |
| channels_[i]->CrossFade(append_this[i], fade_length); |
| } |
| } |
| } |
| |
| template<typename T> |
| size_t AudioMultiVector<T>::Size() const { |
| assert(channels_[0]); |
| return channels_[0]->Size(); |
| } |
| |
| template<typename T> |
| void AudioMultiVector<T>::AssertSize(size_t required_size) { |
| if (Size() < required_size) { |
| size_t extend_length = required_size - Size(); |
| for (size_t channel = 0; channel < Channels(); ++channel) { |
| channels_[channel]->Extend(extend_length); |
| } |
| } |
| } |
| |
| template<typename T> |
| bool AudioMultiVector<T>::Empty() const { |
| assert(channels_[0]); |
| return channels_[0]->Empty(); |
| } |
| |
| template<typename T> |
| const AudioVector<T>& AudioMultiVector<T>::operator[](size_t index) const { |
| return *(channels_[index]); |
| } |
| |
| template<typename T> |
| AudioVector<T>& AudioMultiVector<T>::operator[](size_t index) { |
| return *(channels_[index]); |
| } |
| |
| // Instantiate the template for a few types. |
| template class AudioMultiVector<int16_t>; |
| template class AudioMultiVector<int32_t>; |
| template class AudioMultiVector<double>; |
| |
| } // namespace webrtc |