audio-processor-file 3.3.0

// Augmented Audio: Audio libraries and applications
// Copyright (c) 2022 Pedro Tacla Yamada
//
// The MIT License (MIT)
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.

use std::fs::File;
use std::path::Path;

use symphonia::core::audio::Signal;
use symphonia::core::audio::{AudioBuffer as SymphoniaAudioBuffer, AudioBufferRef};
use symphonia::core::codecs::Decoder;
use symphonia::core::formats::FormatOptions;
use symphonia::core::io::MediaSourceStream;
use symphonia::core::meta::MetadataOptions;
use symphonia::core::probe::{Hint, ProbeResult};
use symphonia::default::get_probe;

use audio_processor_traits::AudioBuffer;
use augmented_audio_metrics as metrics;

use crate::file_io::sample_rate_converter::BLOCK_SIZE;

pub use self::audio_file_error::AudioFileError;

mod audio_file_error;
mod sample_rate_converter;
#[cfg(test)]
mod test;

/// Opens an audio file with default options & trying to guess the format
pub fn default_read_audio_file(input_audio_path: &str) -> Result<ProbeResult, AudioFileError> {
    log::info!(
        "Trying to open and probe audio file at {}",
        input_audio_path
    );

    let mut hint = Hint::new();
    let media_source = {
        let audio_input_path = Path::new(input_audio_path);
        let _ = try_set_audio_file_hint(&mut hint, audio_input_path);
        File::open(audio_input_path)?
    };
    let audio_file = MediaSourceStream::new(Box::new(media_source), Default::default());
    let format_opts: FormatOptions = Default::default();
    let metadata_opts: MetadataOptions = Default::default();
    let audio_file = get_probe().format(&hint, audio_file, &format_opts, &metadata_opts)?;
    Ok(audio_file)
}

/// Attempt to set a hint on codec based on the input file extension
fn try_set_audio_file_hint(hint: &mut Hint, audio_input_path: &Path) -> Option<()> {
    let extension = audio_input_path.extension()?;
    let extension_str = extension.to_str()?;
    hint.with_extension(extension_str);
    Some(())
}

pub struct FileContentsStream<'a> {
    audio_file: &'a mut ProbeResult,
    decoder: Box<dyn Decoder>,
    audio_file_stream_id: u32,
}

impl<'a> FileContentsStream<'a> {
    pub fn new(audio_file: &'a mut ProbeResult) -> Result<Self, AudioFileError> {
        let audio_file_stream = audio_file
            .format
            .default_track()
            .ok_or(AudioFileError::OpenStreamError)?;
        let decoder = symphonia::default::get_codecs()
            .make(&audio_file_stream.codec_params, &Default::default())?;
        let audio_file_stream_id = audio_file_stream.id;

        Ok(FileContentsStream {
            audio_file,
            audio_file_stream_id,
            decoder,
        })
    }
}

impl<'a> Iterator for FileContentsStream<'a> {
    type Item = SymphoniaAudioBuffer<f32>;

    fn next(&mut self) -> Option<SymphoniaAudioBuffer<f32>> {
        let packet = self.audio_file.format.next_packet().ok()?;

        if packet.track_id() != self.audio_file_stream_id {
            return None;
        }

        let audio_buffer = self.decoder.decode(&packet).ok()?;
        let destination = convert_audio_buffer_sample_type(audio_buffer);
        Some(destination)
    }
}

impl<'a> ExactSizeIterator for FileContentsStream<'a> {
    fn len(&self) -> usize {
        self.audio_file
            .format
            .default_track()
            .map(|track| track.codec_params.n_frames.unwrap_or(0))
            .unwrap_or(0) as usize
    }
}

pub fn read_file_contents(
    audio_file: &mut ProbeResult,
) -> Result<SymphoniaAudioBuffer<f32>, AudioFileError> {
    let stream = FileContentsStream::new(audio_file)?;

    let mut channel_buffers: Vec<SymphoniaAudioBuffer<f32>> = Vec::new();
    for buffer in stream {
        channel_buffers.push(buffer)
    }

    if channel_buffers.is_empty() {
        return Err(AudioFileError::EmptyFileError);
    }

    Ok(metrics::time(
        "AudioFileProcessor - Concatenating packets",
        || concat_buffers(channel_buffers),
    ))
}

pub fn convert_audio_buffer_sample_type(audio_buffer: AudioBufferRef) -> SymphoniaAudioBuffer<f32> {
    let mut destination =
        SymphoniaAudioBuffer::new(audio_buffer.capacity() as u64, *audio_buffer.spec());
    let _ = destination.fill(|_, _| Ok(()));
    destination.truncate(audio_buffer.frames());
    assert_eq!(audio_buffer.frames(), destination.frames());
    assert_eq!(audio_buffer.capacity(), destination.capacity());

    match audio_buffer {
        AudioBufferRef::U8(inner) => inner.convert(&mut destination),
        AudioBufferRef::U16(inner) => inner.convert(&mut destination),
        AudioBufferRef::U24(inner) => inner.convert(&mut destination),
        AudioBufferRef::U32(inner) => inner.convert(&mut destination),
        AudioBufferRef::S8(inner) => inner.convert(&mut destination),
        AudioBufferRef::S16(inner) => inner.convert(&mut destination),
        AudioBufferRef::S24(inner) => inner.convert(&mut destination),
        AudioBufferRef::S32(inner) => inner.convert(&mut destination),
        AudioBufferRef::F32(inner) => inner.convert(&mut destination),
        AudioBufferRef::F64(inner) => inner.convert(&mut destination),
    }
    destination
}

struct FileFramesStream<BufferIterator> {
    audio_file_stream: BufferIterator,
    buffer: Vec<Vec<f32>>,
    rate: u32,
    buffer_size: usize,
}

impl<BufferIterator> FileFramesStream<BufferIterator> {
    fn new(audio_file_stream: BufferIterator, buffer_size: usize) -> Self {
        Self {
            audio_file_stream,
            buffer: Vec::new(),
            rate: 0,
            buffer_size,
        }
    }
}

impl<BufferIterator> Iterator for FileFramesStream<BufferIterator>
where
    BufferIterator: Iterator<Item = SymphoniaAudioBuffer<f32>>,
{
    type Item = (Vec<Vec<f32>>, usize);

    fn next(&mut self) -> Option<Self::Item> {
        for block in self.audio_file_stream.by_ref() {
            let channels = block.spec().channels.count();
            self.rate = block.spec().rate;
            self.buffer.resize(channels, Vec::new());
            for channel in 0..channels {
                self.buffer[channel].extend_from_slice(block.chan(channel));
            }

            if self.buffer[0].len() >= self.buffer_size {
                let mut result = Vec::new();
                for channel in 0..self.buffer.len() {
                    let channel_buffer: Vec<f32> =
                        self.buffer[channel].drain(0..self.buffer_size).collect();

                    assert_eq!(channel_buffer.len(), self.buffer_size);
                    result.push(channel_buffer);
                }

                return Some((result, self.buffer_size));
            }
        }

        if !self.buffer[0].is_empty() {
            let mut result = Vec::new();
            // On the trailing end of the buffer, this might be smaller than block-size. We need to
            // output a full block, but also let the consumer know this last block is special.
            let mut remaining_length = 0;
            for channel in 0..self.buffer.len() {
                remaining_length = self.buffer_size.min(self.buffer[channel].len());
                let mut channel_buffer: Vec<f32> =
                    self.buffer[channel].drain(0..remaining_length).collect();
                channel_buffer.resize(self.buffer_size, 0.0);
                result.push(channel_buffer);
            }

            assert!(remaining_length > 0);
            assert!(remaining_length <= self.buffer_size);
            assert!(remaining_length <= result[0].len());
            return Some((result, remaining_length));
        }
        None
    }
}

pub struct ConvertedFileContentsStream<'a> {
    audio_file_stream: FileFramesStream<FileContentsStream<'a>>,
    output_rate: f32,
    decoder: Option<sample_rate_converter::Decoder>,
}

impl<'a> ConvertedFileContentsStream<'a> {
    fn get_decoder(
        &mut self,
        from_rate: u32,
        channels: usize,
    ) -> Option<&mut sample_rate_converter::Decoder> {
        if self.decoder.is_none() {
            self.decoder = Some(
                sample_rate_converter::make_decoder(from_rate, self.output_rate as u32, channels)
                    .unwrap(),
            );
        }

        if let Some(decoder) = &mut self.decoder {
            Some(decoder)
        } else {
            None
        }
    }
}

impl<'a> Iterator for ConvertedFileContentsStream<'a> {
    type Item = AudioBuffer<f32>;

    fn next(&mut self) -> Option<Self::Item> {
        let (channels, chunk_size) = self.audio_file_stream.next()?;
        let rate = self.audio_file_stream.rate;

        let decoder = self.get_decoder(rate, channels.len())?;

        assert_eq!(channels.len(), 2);
        assert_eq!(channels[0].len(), BLOCK_SIZE);
        // Resize channels so that the trailing smaller block is resized
        let mut chunk_result = sample_rate_converter::process(decoder, &channels).ok()?;
        let expected_chunk_size = (chunk_size as f32 * self.output_rate / rate as f32) as usize;
        if chunk_size != BLOCK_SIZE {
            for channel in chunk_result.iter_mut() {
                channel.resize(expected_chunk_size, 0.0);
            }
        }

        Some(AudioBuffer::new(chunk_result))
    }
}

pub fn convert_audio_file_stream_sample_rate(
    audio_file_stream: FileContentsStream,
    output_rate: f32,
) -> ConvertedFileContentsStream {
    ConvertedFileContentsStream {
        audio_file_stream: FileFramesStream::new(audio_file_stream, BLOCK_SIZE),
        output_rate,
        decoder: None,
    }
}

#[cfg(feature = "samplerate")]
pub fn convert_audio_file_sample_rate(
    audio_file_contents: &SymphoniaAudioBuffer<f32>,
    output_rate: f32,
    channel_number: usize,
) -> Vec<f32> {
    let audio_file_channel = audio_file_contents.chan(channel_number);
    let input_rate = audio_file_contents.spec().rate as f32;
    let audio_file_duration = audio_file_channel.len() as f32 / input_rate;

    let output_size = (audio_file_duration * output_rate).ceil() as usize;
    let mut channel = Vec::new();
    channel.resize(output_size, 0.0);

    // Convert sample rate from audio file to in-memory
    log::info!(
        "Converting sample_rate channel={} input_rate={} output_rate={}",
        channel_number,
        input_rate,
        output_rate
    );
    augmented_convert_sample_rate::convert_sample_rate(
        input_rate,
        audio_file_channel,
        output_rate,
        channel.as_mut_slice(),
    );

    channel
}

/// buffers must be non-empty and stereo
fn concat_buffers(buffers: Vec<SymphoniaAudioBuffer<f32>>) -> SymphoniaAudioBuffer<f32> {
    let duration = buffers
        .iter()
        .map(|buffer| buffer.chan(0).len() as u64)
        .sum();

    let mut output = SymphoniaAudioBuffer::new(duration, *buffers[0].spec());
    let _ = output.fill(|_, _| Ok(()));
    let mut output_cursor = 0;
    for buffer in buffers {
        let mut channel_size = 0;

        for channel_num in 0..2 {
            let mut cursor = output_cursor; // reading channels copy cursor to reset for each channel

            let output_channel = output.chan_mut(channel_num);
            let channel = buffer.chan(channel_num);
            channel_size = channel.len();

            for sample in channel {
                output_channel[cursor] = *sample;
                cursor += 1;
            }
        }

        output_cursor += channel_size;
    }
    output
}