AudioResamplerTemplate.cpp

/******************************************************************************
    QtAV:  Multimedia framework based on Qt and FFmpeg
    Copyright (C) 2012-2016 Wang Bin <wbsecg1@gmail.com>

*   This file is part of QtAV

    This library 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.

    This library 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 this library; if not, write to the Free Software
    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
******************************************************************************/

#if defined(BUILD_AVR) || defined(BUILD_SWR) // no this macro is fine too for qmake
#include "QtAV/AudioResampler.h"
#include "QtAV/private/AudioResampler_p.h"
#include "QtAV/private/AVCompat.h"
#include "QtAV/private/factory.h"
#include "utils/Logger.h"

namespace QtAV {

#if QTAV_HAVE(SWR_AVR_MAP)
#define AudioResamplerFF AudioResamplerLibav
#define AudioResamplerFFPrivate AudioResamplerLibavPrivate
#define AudioResamplerId_FF AudioResamplerId_Libav
#define RegisterAudioResamplerFF_Man RegisterAudioResamplerLibav_Man
#define FF Libav
static const char kName[] = "Libav";
#else
static const char kName[] = "FFmpeg";
#endif

class AudioResamplerFFPrivate;
class AudioResamplerFF : public AudioResampler
{
    DPTR_DECLARE_PRIVATE(AudioResampler)
public:
    AudioResamplerFF();
    virtual bool convert(const quint8** data);
    virtual bool prepare();
};
extern AudioResamplerId AudioResamplerId_FF;
FACTORY_REGISTER(AudioResampler, FF, kName)

class AudioResamplerFFPrivate : public AudioResamplerPrivate
{
public:
    AudioResamplerFFPrivate(): context(0) {}
    ~AudioResamplerFFPrivate() {
        if (context) {
            swr_free(&context);
            context = 0;
        }
    }
    SwrContext *context;
    // defined in swr<1
#ifndef SWR_CH_MAX
#define SWR_CH_MAX 64
#endif
    int channel_map[SWR_CH_MAX];
};

AudioResamplerFF::AudioResamplerFF():
    AudioResampler(*new AudioResamplerFFPrivate())
{
}

bool AudioResamplerFF::convert(const quint8 **data)
{
    DPTR_D(AudioResamplerFF);
    /*
     * swr_get_delay: Especially when downsampling by a large value, the output sample rate may be a poor choice to represent
     * the delay, similarly  upsampling and the input sample rate.
     */
    qreal osr = d.out_format.sampleRate();
    if (!qFuzzyCompare(d.speed, 1.0))
        osr /= d.speed;
    d.out_samples_per_channel = av_rescale_rnd(
#if HAVE_SWR_GET_DELAY
                swr_get_delay(d.context, qMax(d.in_format.sampleRate(), d.out_format.sampleRate())) +
#else
                128 + //TODO: QtAV_Compat
#endif //HAVE_SWR_GET_DELAY
                d.in_samples_per_channel //TODO: wanted_samples(ffplay mplayer2)
                , osr, d.in_format.sampleRate(), AV_ROUND_UP);
    //TODO: why crash for swr 0.5?
    //int out_size = av_samples_get_buffer_size(NULL/*out linesize*/, d.out_channels, d.out_samples_per_channel, (AVSampleFormat)d.out_sample_format, 0/*alignment default*/);
    int size_per_sample_with_channels = d.out_format.channels()*d.out_format.bytesPerSample();
    int out_size = d.out_samples_per_channel*size_per_sample_with_channels;
    if (out_size > d.data_out.size())
        d.data_out.resize(out_size);
    uint8_t *out[] = {(uint8_t*)d.data_out.data()}; // detach if implicitly shared by others
    //number of input/output samples available in one channel
    int converted_samplers_per_channel = swr_convert(d.context, out, d.out_samples_per_channel, data, d.in_samples_per_channel);
    d.out_samples_per_channel = converted_samplers_per_channel;
    if (converted_samplers_per_channel < 0) {
        qWarning("[AudioResamplerFF] %s", av_err2str(converted_samplers_per_channel));
        return false;
    }
    //TODO: converted_samplers_per_channel==out_samples_per_channel means out_size is too small, see mplayer2
    //converted_samplers_per_channel*d.out_channels*av_get_bytes_per_sample(d.out_sample_format)
    //av_samples_get_buffer_size(0, d.out_channels, converted_samplers_per_channel, d.out_sample_format, 0)
    //if (converted_samplers_per_channel != out_size)
    d.data_out.resize(converted_samplers_per_channel*size_per_sample_with_channels);
    return true;
}

/*
 *TODO: broken sample rate(AAC), see mplayer
 */
bool AudioResamplerFF::prepare()
{
    DPTR_D(AudioResamplerFF);
    if (!d.in_format.isValid()) {
        qWarning("src audio parameters 'channel layout(or channels), sample rate and sample format must be set before initialize resampler");
        return false;
    }
    //TODO: also in do this statistics
    if (!d.in_format.channels()) {
        if (!d.in_format.channelLayoutFFmpeg()) { //FIXME: already return
            d.in_format.setChannels(2);
            d.in_format.setChannelLayoutFFmpeg(av_get_default_channel_layout(d.in_format.channels())); //from mplayer2
            qWarning("both channels and channel layout are not available, assume channels=%d, channel layout=%lld", d.in_format.channels(), d.in_format.channelLayoutFFmpeg());
        } else {
            d.in_format.setChannels(av_get_channel_layout_nb_channels(d.in_format.channelLayoutFFmpeg()));
        }
    }
    if (!d.in_format.channels())
        d.in_format.setChannels(2); //TODO: why av_get_channel_layout_nb_channels() may return 0?
    if (!d.in_format.channelLayoutFFmpeg()) {
        qWarning("channel layout not available, use default layout");
        d.in_format.setChannelLayoutFFmpeg(av_get_default_channel_layout(d.in_format.channels()));
    }
    if (!d.out_format.channels()) {
        if (d.out_format.channelLayoutFFmpeg()) {
            d.out_format.setChannels(av_get_channel_layout_nb_channels(d.out_format.channelLayoutFFmpeg()));
        } else {
            d.out_format.setChannels(d.in_format.channels());
            d.out_format.setChannelLayoutFFmpeg(d.in_format.channelLayoutFFmpeg());
        }
    }
    if (d.out_format.channelLayout() == AudioFormat::ChannelLayout_Unsupported) {
        d.out_format.setChannels(d.in_format.channels());
        d.out_format.setChannelLayoutFFmpeg(d.in_format.channelLayoutFFmpeg());
    }
    //now we have out channels
    if (!d.out_format.channelLayoutFFmpeg())
        d.out_format.setChannelLayoutFFmpeg(av_get_default_channel_layout(d.out_format.channels()));
    if (!d.out_format.sampleRate())
        d.out_format.setSampleRate(inAudioFormat().sampleRate());
    if (d.speed <= 0)
        d.speed = 1.0;
    //DO NOT set sample rate here, we should keep the original and multiply 1/speed when needed
    //if (d.speed != 1.0)
    //    d.out_format.setSampleRate(int(qreal(d.out_format.sampleFormat())/d.speed));
    qDebug("swr speed=%.2f", d.speed);

    //d.in_planes = av_sample_fmt_is_planar((enum AVSampleFormat)d.in_sample_format) ? d.in_channels : 1;
    //d.out_planes = av_sample_fmt_is_planar((enum AVSampleFormat)d.out_sample_format) ? d.out_channels : 1;
    if (d.context)
        swr_free(&d.context); //TODO: if no free(of cause free is required), why channel mapping and layout not work if change from left to stereo?
    //If use swr_alloc() need to set the parameters (av_opt_set_xxx() manually or with swr_alloc_set_opts()) before calling swr_init()
    d.context = swr_alloc_set_opts(d.context
                                   , d.out_format.channelLayoutFFmpeg()
                                   , (enum AVSampleFormat)outAudioFormat().sampleFormatFFmpeg()
                                   , qreal(outAudioFormat().sampleRate())/d.speed
                                   , d.in_format.channelLayoutFFmpeg()
                                   , (enum AVSampleFormat)inAudioFormat().sampleFormatFFmpeg()
                                   , inAudioFormat().sampleRate()
                                   , 0 /*log_offset*/, 0 /*log_ctx*/);
    /*
    av_opt_set_int(d.context, "in_channel_layout",    d.in_channel_layout, 0);
    av_opt_set_int(d.context, "in_sample_rate",       d.in_format.sampleRate(), 0);
    av_opt_set_sample_fmt(d.context, "in_sample_fmt", (enum AVSampleFormat)in_format.sampleFormatFFmpeg(), 0);
    av_opt_set_int(d.context, "out_channel_layout",    d.out_channel_layout, 0);
    av_opt_set_int(d.context, "out_sample_rate",       d.out_format.sampleRate(), 0);
    av_opt_set_sample_fmt(d.context, "out_sample_fmt", (enum AVSampleFormat)out_format.sampleFormatFFmpeg(), 0);
    */
    qDebug("out: {cl: %lld, fmt: %s, freq: %d}"
           , d.out_format.channelLayoutFFmpeg()
           , qPrintable(d.out_format.sampleFormatName())
           , d.out_format.sampleRate());
    qDebug("in {cl: %lld, fmt: %s, freq: %d}"
           , d.in_format.channelLayoutFFmpeg()
           , qPrintable(d.in_format.sampleFormatName())
           , d.in_format.sampleRate());

    if (!d.context) {
        qWarning("Allocat swr context failed!");
        return false;
    }
    //avresample 0.0.2(FFmpeg 0.11)~1.0.1(FFmpeg 1.1) has no channel mapping. but has remix matrix, so does swresample
//TODO: why crash if use channel mapping for L or R?
#if QTAV_HAVE(SWR_AVR_MAP) //LIBAVRESAMPLE_VERSION_INT < AV_VERSION_INT(1, 1, 0)
    bool remix = false;
    int in_c = d.in_format.channels();
    int out_c = d.out_format.channels();
    /*
     * matrix[i + stride * o] is the weight of input channel i in output channel o.
     */
    double *matrix = 0;
    if (d.out_format.channelLayout() == AudioFormat::ChannelLayout_Left) {
        remix = true;
        matrix = (double*)calloc(in_c*out_c, sizeof(double));
        for (int o = 0; o < out_c; ++o) {
            matrix[0 + in_c * o] = 1;
        }
    }
    if (d.out_format.channelLayout() == AudioFormat::ChannelLayout_Right) {
        remix = true;
        matrix = (double*)calloc(in_c*out_c, sizeof(double));
        for (int o = 0; o < out_c; ++o) {
            matrix[1 + in_c * o] = 1;
        }
    }
    if (!remix && in_c < out_c) {
        remix = true;
        //double matrix[in_c*out_c]; //C99, VLA
        matrix = (double*)calloc(in_c*out_c, sizeof(double));
        for (int i = 0, o = 0; o < out_c; ++o) {
            matrix[i + in_c * o] = 1;
            i = (i + i)%in_c;
        }
    }
    if (remix && matrix) {
        avresample_set_matrix(d.context, matrix, in_c);
        free(matrix);
    }
#else
    bool use_channel_map = false;
    if (d.out_format.channelLayout() == AudioFormat::ChannelLayout_Left) {
        use_channel_map = true;
        memset(d.channel_map, 0, sizeof(d.channel_map));
        for (int i = 0; i < d.out_format.channels(); ++i) {
            d.channel_map[i] = 0;
        }
    }
    if (d.out_format.channelLayout() == AudioFormat::ChannelLayout_Right) {
        use_channel_map = true;
        memset(d.channel_map, 0, sizeof(d.channel_map));
        for (int i = 0; i < d.out_format.channels(); ++i) {
            d.channel_map[i] = 1;
        }
    }
    if (!use_channel_map && d.in_format.channels() < d.out_format.channels()) {
        use_channel_map = true;
        memset(d.channel_map, 0, sizeof(d.channel_map));
        for (int i = 0; i < d.out_format.channels(); ++i) {
            d.channel_map[i] = i % d.in_format.channels();
        }
    }
    if (use_channel_map) {
        av_opt_set_int(d.context, "icl", d.out_format.channelLayoutFFmpeg(), 0);
        //TODO: why crash if layout is mono and set uch(i.e. always the next line)
        av_opt_set_int(d.context, "uch", d.out_format.channels(), 0);
        swr_set_channel_mapping(d.context, d.channel_map);
    }
#endif //QTAV_HAVE(SWR_AVR_MAP)
    int ret = swr_init(d.context);
    if (ret < 0) {
        qWarning("swr_init failed: %s", av_err2str(ret));
        swr_free(&d.context);
        return false;
    }
    return true;
}

} //namespace QtAV
#endif //defined(BUILD_AVR) || defined(BUILD_SWR)