FFmpeg coverage

GCC Code Coverage Report

Directory:	../../../ffmpeg/		Exec	Total	Coverage
File:	src/libavfilter/af_amix.c	Lines:	237	294	80.6 %
Date:	2021-04-14 23:45:22	Branches:	136	192	70.8 %


/*
 * Audio Mix Filter
 * Copyright (c) 2012 Justin Ruggles <justin.ruggles@gmail.com>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg 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.
 *
 * FFmpeg 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 FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
 * @file
 * Audio Mix Filter
 *
 * Mixes audio from multiple sources into a single output. The channel layout,
 * sample rate, and sample format will be the same for all inputs and the
 * output.
 */

#include "libavutil/attributes.h"
#include "libavutil/audio_fifo.h"
#include "libavutil/avassert.h"
#include "libavutil/avstring.h"
#include "libavutil/channel_layout.h"
#include "libavutil/common.h"
#include "libavutil/eval.h"
#include "libavutil/float_dsp.h"
#include "libavutil/mathematics.h"
#include "libavutil/opt.h"
#include "libavutil/samplefmt.h"

#include "audio.h"
#include "avfilter.h"
#include "filters.h"
#include "formats.h"
#include "internal.h"

#define INPUT_ON       1    /**< input is active */
#define INPUT_EOF      2    /**< input has reached EOF (may still be active) */

#define DURATION_LONGEST  0
#define DURATION_SHORTEST 1
#define DURATION_FIRST    2


typedef struct FrameInfo {
    int nb_samples;
    int64_t pts;
    struct FrameInfo *next;
} FrameInfo;

/**
 * Linked list used to store timestamps and frame sizes of all frames in the
 * FIFO for the first input.
 *
 * This is needed to keep timestamps synchronized for the case where multiple
 * input frames are pushed to the filter for processing before a frame is
 * requested by the output link.
 */
typedef struct FrameList {
    int nb_frames;
    int nb_samples;
    FrameInfo *list;
    FrameInfo *end;
} FrameList;

static void frame_list_clear(FrameList *frame_list)
{
    if (frame_list) {
        while (frame_list->list) {
            FrameInfo *info = frame_list->list;
            frame_list->list = info->next;
            av_free(info);
        }
        frame_list->nb_frames  = 0;
        frame_list->nb_samples = 0;
        frame_list->end        = NULL;
    }
}

static int frame_list_next_frame_size(FrameList *frame_list)
{
    if (!frame_list->list)
        return 0;
    return frame_list->list->nb_samples;
}

static int64_t frame_list_next_pts(FrameList *frame_list)
{
    if (!frame_list->list)
        return AV_NOPTS_VALUE;
    return frame_list->list->pts;
}

static void frame_list_remove_samples(FrameList *frame_list, int nb_samples)
{
    if (nb_samples >= frame_list->nb_samples) {
        frame_list_clear(frame_list);
    } else {
        int samples = nb_samples;
        while (samples > 0) {
            FrameInfo *info = frame_list->list;
            av_assert0(info);
            if (info->nb_samples <= samples) {
                samples -= info->nb_samples;
                frame_list->list = info->next;
                if (!frame_list->list)
                    frame_list->end = NULL;
                frame_list->nb_frames--;
                frame_list->nb_samples -= info->nb_samples;
                av_free(info);
            } else {
                info->nb_samples       -= samples;
                info->pts              += samples;
                frame_list->nb_samples -= samples;
                samples = 0;
            }
        }
    }
}

static int frame_list_add_frame(FrameList *frame_list, int nb_samples, int64_t pts)
{
    FrameInfo *info = av_malloc(sizeof(*info));
    if (!info)
        return AVERROR(ENOMEM);
    info->nb_samples = nb_samples;
    info->pts        = pts;
    info->next       = NULL;

    if (!frame_list->list) {
        frame_list->list = info;
        frame_list->end  = info;
    } else {
        av_assert0(frame_list->end);
        frame_list->end->next = info;
        frame_list->end       = info;
    }
    frame_list->nb_frames++;
    frame_list->nb_samples += nb_samples;

    return 0;
}

/* FIXME: use directly links fifo */

typedef struct MixContext {
    const AVClass *class;       /**< class for AVOptions */
    AVFloatDSPContext *fdsp;

    int nb_inputs;              /**< number of inputs */
    int active_inputs;          /**< number of input currently active */
    int duration_mode;          /**< mode for determining duration */
    float dropout_transition;   /**< transition time when an input drops out */
    char *weights_str;          /**< string for custom weights for every input */
    int normalize;              /**< if inputs are scaled */

    int nb_channels;            /**< number of channels */
    int sample_rate;            /**< sample rate */
    int planar;
    AVAudioFifo **fifos;        /**< audio fifo for each input */
    uint8_t *input_state;       /**< current state of each input */
    float *input_scale;         /**< mixing scale factor for each input */
    float *weights;             /**< custom weights for every input */
    float weight_sum;           /**< sum of custom weights for every input */
    float *scale_norm;          /**< normalization factor for every input */
    int64_t next_pts;           /**< calculated pts for next output frame */
    FrameList *frame_list;      /**< list of frame info for the first input */
} MixContext;

#define OFFSET(x) offsetof(MixContext, x)
#define A AV_OPT_FLAG_AUDIO_PARAM
#define F AV_OPT_FLAG_FILTERING_PARAM
#define T AV_OPT_FLAG_RUNTIME_PARAM
static const AVOption amix_options[] = {
    { "inputs", "Number of inputs.",
            OFFSET(nb_inputs), AV_OPT_TYPE_INT, { .i64 = 2 }, 1, INT16_MAX, A|F },
    { "duration", "How to determine the end-of-stream.",
            OFFSET(duration_mode), AV_OPT_TYPE_INT, { .i64 = DURATION_LONGEST }, 0,  2, A|F, "duration" },
        { "longest",  "Duration of longest input.",  0, AV_OPT_TYPE_CONST, { .i64 = DURATION_LONGEST  }, 0, 0, A|F, "duration" },
        { "shortest", "Duration of shortest input.", 0, AV_OPT_TYPE_CONST, { .i64 = DURATION_SHORTEST }, 0, 0, A|F, "duration" },
        { "first",    "Duration of first input.",    0, AV_OPT_TYPE_CONST, { .i64 = DURATION_FIRST    }, 0, 0, A|F, "duration" },
    { "dropout_transition", "Transition time, in seconds, for volume "
                            "renormalization when an input stream ends.",
            OFFSET(dropout_transition), AV_OPT_TYPE_FLOAT, { .dbl = 2.0 }, 0, INT_MAX, A|F },
    { "weights", "Set weight for each input.",
            OFFSET(weights_str), AV_OPT_TYPE_STRING, {.str="1 1"}, 0, 0, A|F|T },
    { "normalize", "Scale inputs",
            OFFSET(normalize), AV_OPT_TYPE_BOOL, {.i64=1}, 0, 1, A|F|T },
    { NULL }
};

AVFILTER_DEFINE_CLASS(amix);

/**
 * Update the scaling factors to apply to each input during mixing.
 *
 * This balances the full volume range between active inputs and handles
 * volume transitions when EOF is encountered on an input but mixing continues
 * with the remaining inputs.
 */
static void calculate_scales(MixContext *s, int nb_samples)
{
    float weight_sum = 0.f;
    int i;

    for (i = 0; i < s->nb_inputs; i++)
        if (s->input_state[i] & INPUT_ON)
            weight_sum += FFABS(s->weights[i]);

    for (i = 0; i < s->nb_inputs; i++) {
        if (s->input_state[i] & INPUT_ON) {
            if (s->scale_norm[i] > weight_sum / FFABS(s->weights[i])) {
                s->scale_norm[i] -= ((s->weight_sum / FFABS(s->weights[i])) / s->nb_inputs) *
                                    nb_samples / (s->dropout_transition * s->sample_rate);
                s->scale_norm[i] = FFMAX(s->scale_norm[i], weight_sum / FFABS(s->weights[i]));
            }
        }
    }

    for (i = 0; i < s->nb_inputs; i++) {
        if (s->input_state[i] & INPUT_ON) {
            if (!s->normalize)
                s->input_scale[i] = FFABS(s->weights[i]);
            else
                s->input_scale[i] = 1.0f / s->scale_norm[i] * FFSIGN(s->weights[i]);
        } else {
            s->input_scale[i] = 0.0f;
        }
    }
}

static int config_output(AVFilterLink *outlink)
{
    AVFilterContext *ctx = outlink->src;
    MixContext *s      = ctx->priv;
    int i;
    char buf[64];

    s->planar          = av_sample_fmt_is_planar(outlink->format);
    s->sample_rate     = outlink->sample_rate;
    outlink->time_base = (AVRational){ 1, outlink->sample_rate };
    s->next_pts        = AV_NOPTS_VALUE;

    s->frame_list = av_mallocz(sizeof(*s->frame_list));
    if (!s->frame_list)
        return AVERROR(ENOMEM);

    s->fifos = av_mallocz_array(s->nb_inputs, sizeof(*s->fifos));
    if (!s->fifos)
        return AVERROR(ENOMEM);

    s->nb_channels = outlink->channels;
    for (i = 0; i < s->nb_inputs; i++) {
        s->fifos[i] = av_audio_fifo_alloc(outlink->format, s->nb_channels, 1024);
        if (!s->fifos[i])
            return AVERROR(ENOMEM);
    }

    s->input_state = av_malloc(s->nb_inputs);
    if (!s->input_state)
        return AVERROR(ENOMEM);
    memset(s->input_state, INPUT_ON, s->nb_inputs);
    s->active_inputs = s->nb_inputs;

    s->input_scale = av_mallocz_array(s->nb_inputs, sizeof(*s->input_scale));
    s->scale_norm  = av_mallocz_array(s->nb_inputs, sizeof(*s->scale_norm));
    if (!s->input_scale || !s->scale_norm)
        return AVERROR(ENOMEM);
    for (i = 0; i < s->nb_inputs; i++)
        s->scale_norm[i] = s->weight_sum / FFABS(s->weights[i]);
    calculate_scales(s, 0);

    av_get_channel_layout_string(buf, sizeof(buf), -1, outlink->channel_layout);

    av_log(ctx, AV_LOG_VERBOSE,
           "inputs:%d fmt:%s srate:%d cl:%s\n", s->nb_inputs,
           av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf);

    return 0;
}

/**
 * Read samples from the input FIFOs, mix, and write to the output link.
 */
static int output_frame(AVFilterLink *outlink)
{
    AVFilterContext *ctx = outlink->src;
    MixContext      *s = ctx->priv;
    AVFrame *out_buf, *in_buf;
    int nb_samples, ns, i;

    if (s->input_state[0] & INPUT_ON) {
        /* first input live: use the corresponding frame size */
        nb_samples = frame_list_next_frame_size(s->frame_list);
        for (i = 1; i < s->nb_inputs; i++) {
            if (s->input_state[i] & INPUT_ON) {
                ns = av_audio_fifo_size(s->fifos[i]);
                if (ns < nb_samples) {
                    if (!(s->input_state[i] & INPUT_EOF))
                        /* unclosed input with not enough samples */
                        return 0;
                    /* closed input to drain */
                    nb_samples = ns;
                }
            }
        }

        s->next_pts = frame_list_next_pts(s->frame_list);
    } else {
        /* first input closed: use the available samples */
        nb_samples = INT_MAX;
        for (i = 1; i < s->nb_inputs; i++) {
            if (s->input_state[i] & INPUT_ON) {
                ns = av_audio_fifo_size(s->fifos[i]);
                nb_samples = FFMIN(nb_samples, ns);
            }
        }
        if (nb_samples == INT_MAX) {
            ff_outlink_set_status(outlink, AVERROR_EOF, s->next_pts);
            return 0;
        }
    }

    frame_list_remove_samples(s->frame_list, nb_samples);

    calculate_scales(s, nb_samples);

    if (nb_samples == 0)
        return 0;

    out_buf = ff_get_audio_buffer(outlink, nb_samples);
    if (!out_buf)
        return AVERROR(ENOMEM);

    in_buf = ff_get_audio_buffer(outlink, nb_samples);
    if (!in_buf) {
        av_frame_free(&out_buf);
        return AVERROR(ENOMEM);
    }

    for (i = 0; i < s->nb_inputs; i++) {
        if (s->input_state[i] & INPUT_ON) {
            int planes, plane_size, p;

            av_audio_fifo_read(s->fifos[i], (void **)in_buf->extended_data,
                               nb_samples);

            planes     = s->planar ? s->nb_channels : 1;
            plane_size = nb_samples * (s->planar ? 1 : s->nb_channels);
            plane_size = FFALIGN(plane_size, 16);

            if (out_buf->format == AV_SAMPLE_FMT_FLT ||
                out_buf->format == AV_SAMPLE_FMT_FLTP) {
                for (p = 0; p < planes; p++) {
                    s->fdsp->vector_fmac_scalar((float *)out_buf->extended_data[p],
                                                (float *) in_buf->extended_data[p],
                                                s->input_scale[i], plane_size);
                }
            } else {
                for (p = 0; p < planes; p++) {
                    s->fdsp->vector_dmac_scalar((double *)out_buf->extended_data[p],
                                                (double *) in_buf->extended_data[p],
                                                s->input_scale[i], plane_size);
                }
            }
        }
    }
    av_frame_free(&in_buf);

    out_buf->pts = s->next_pts;
    if (s->next_pts != AV_NOPTS_VALUE)
        s->next_pts += nb_samples;

    return ff_filter_frame(outlink, out_buf);
}

/**
 * Requests a frame, if needed, from each input link other than the first.
 */
static int request_samples(AVFilterContext *ctx, int min_samples)
{
    MixContext *s = ctx->priv;
    int i;

    av_assert0(s->nb_inputs > 1);

    for (i = 1; i < s->nb_inputs; i++) {
        if (!(s->input_state[i] & INPUT_ON) ||
             (s->input_state[i] & INPUT_EOF))
            continue;
        if (av_audio_fifo_size(s->fifos[i]) >= min_samples)
            continue;
        ff_inlink_request_frame(ctx->inputs[i]);
    }
    return output_frame(ctx->outputs[0]);
}

/**
 * Calculates the number of active inputs and determines EOF based on the
 * duration option.
 *
 * @return 0 if mixing should continue, or AVERROR_EOF if mixing should stop.
 */
static int calc_active_inputs(MixContext *s)
{
    int i;
    int active_inputs = 0;
    for (i = 0; i < s->nb_inputs; i++)
        active_inputs += !!(s->input_state[i] & INPUT_ON);
    s->active_inputs = active_inputs;

    if (!active_inputs ||
        (s->duration_mode == DURATION_FIRST && !(s->input_state[0] & INPUT_ON)) ||
        (s->duration_mode == DURATION_SHORTEST && active_inputs != s->nb_inputs))
        return AVERROR_EOF;
    return 0;
}

static int activate(AVFilterContext *ctx)
{
    AVFilterLink *outlink = ctx->outputs[0];
    MixContext *s = ctx->priv;
    AVFrame *buf = NULL;
    int i, ret;

    FF_FILTER_FORWARD_STATUS_BACK_ALL(outlink, ctx);

    for (i = 0; i < s->nb_inputs; i++) {
        AVFilterLink *inlink = ctx->inputs[i];

        if ((ret = ff_inlink_consume_frame(ctx->inputs[i], &buf)) > 0) {
            if (i == 0) {
                int64_t pts = av_rescale_q(buf->pts, inlink->time_base,
                                           outlink->time_base);
                ret = frame_list_add_frame(s->frame_list, buf->nb_samples, pts);
                if (ret < 0) {
                    av_frame_free(&buf);
                    return ret;
                }
            }

            ret = av_audio_fifo_write(s->fifos[i], (void **)buf->extended_data,
                                      buf->nb_samples);
            if (ret < 0) {
                av_frame_free(&buf);
                return ret;
            }

            av_frame_free(&buf);

            ret = output_frame(outlink);
            if (ret < 0)
                return ret;
        }
    }

    for (i = 0; i < s->nb_inputs; i++) {
        int64_t pts;
        int status;

        if (ff_inlink_acknowledge_status(ctx->inputs[i], &status, &pts)) {
            if (status == AVERROR_EOF) {
                if (i == 0) {
                    s->input_state[i] = 0;
                    if (s->nb_inputs == 1) {
                        ff_outlink_set_status(outlink, status, pts);
                        return 0;
                    }
                } else {
                    s->input_state[i] |= INPUT_EOF;
                    if (av_audio_fifo_size(s->fifos[i]) == 0) {
                        s->input_state[i] = 0;
                    }
                }
            }
        }
    }

    if (calc_active_inputs(s)) {
        ff_outlink_set_status(outlink, AVERROR_EOF, s->next_pts);
        return 0;
    }

    if (ff_outlink_frame_wanted(outlink)) {
        int wanted_samples;

        if (!(s->input_state[0] & INPUT_ON))
            return request_samples(ctx, 1);

        if (s->frame_list->nb_frames == 0) {
            ff_inlink_request_frame(ctx->inputs[0]);
            return 0;
        }
        av_assert0(s->frame_list->nb_frames > 0);

        wanted_samples = frame_list_next_frame_size(s->frame_list);

        return request_samples(ctx, wanted_samples);
    }

    return 0;
}

static void parse_weights(AVFilterContext *ctx)
{
    MixContext *s = ctx->priv;
    float last_weight = 1.f;
    char *p;
    int i;

    s->weight_sum = 0.f;
    p = s->weights_str;
    for (i = 0; i < s->nb_inputs; i++) {
        last_weight = av_strtod(p, &p);
        s->weights[i] = last_weight;
        s->weight_sum += FFABS(last_weight);
        if (p && *p) {
            p++;
        } else {
            i++;
            break;
        }
    }

    for (; i < s->nb_inputs; i++) {
        s->weights[i] = last_weight;
        s->weight_sum += FFABS(last_weight);
    }
}

static av_cold int init(AVFilterContext *ctx)
{
    MixContext *s = ctx->priv;
    int i, ret;

    for (i = 0; i < s->nb_inputs; i++) {
        AVFilterPad pad = { 0 };

        pad.type           = AVMEDIA_TYPE_AUDIO;
        pad.name           = av_asprintf("input%d", i);
        if (!pad.name)
            return AVERROR(ENOMEM);

        if ((ret = ff_insert_inpad(ctx, i, &pad)) < 0) {
            av_freep(&pad.name);
            return ret;
        }
    }

    s->fdsp = avpriv_float_dsp_alloc(0);
    if (!s->fdsp)
        return AVERROR(ENOMEM);

    s->weights = av_mallocz_array(s->nb_inputs, sizeof(*s->weights));
    if (!s->weights)
        return AVERROR(ENOMEM);

    parse_weights(ctx);

    return 0;
}

static av_cold void uninit(AVFilterContext *ctx)
{
    int i;
    MixContext *s = ctx->priv;

    if (s->fifos) {
        for (i = 0; i < s->nb_inputs; i++)
            av_audio_fifo_free(s->fifos[i]);
        av_freep(&s->fifos);
    }
    frame_list_clear(s->frame_list);
    av_freep(&s->frame_list);
    av_freep(&s->input_state);
    av_freep(&s->input_scale);
    av_freep(&s->scale_norm);
    av_freep(&s->weights);
    av_freep(&s->fdsp);

    for (i = 0; i < ctx->nb_inputs; i++)
        av_freep(&ctx->input_pads[i].name);
}

static int query_formats(AVFilterContext *ctx)
{
    static const enum AVSampleFormat sample_fmts[] = {
        AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
        AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBLP,
        AV_SAMPLE_FMT_NONE
    };
    int ret;

    if ((ret = ff_set_common_formats(ctx, ff_make_format_list(sample_fmts))) < 0 ||
        (ret = ff_set_common_samplerates(ctx, ff_all_samplerates())) < 0)
        return ret;

    return ff_set_common_channel_layouts(ctx, ff_all_channel_counts());
}

static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
                           char *res, int res_len, int flags)
{
    MixContext *s = ctx->priv;
    int ret;

    ret = ff_filter_process_command(ctx, cmd, args, res, res_len, flags);
    if (ret < 0)
        return ret;

    parse_weights(ctx);
    for (int i = 0; i < s->nb_inputs; i++)
        s->scale_norm[i] = s->weight_sum / FFABS(s->weights[i]);
    calculate_scales(s, 0);

    return 0;
}

static const AVFilterPad avfilter_af_amix_outputs[] = {
    {
        .name          = "default",
        .type          = AVMEDIA_TYPE_AUDIO,
        .config_props  = config_output,
    },
    { NULL }
};

AVFilter ff_af_amix = {
    .name           = "amix",
    .description    = NULL_IF_CONFIG_SMALL("Audio mixing."),
    .priv_size      = sizeof(MixContext),
    .priv_class     = &amix_class,
    .init           = init,
    .uninit         = uninit,
    .activate       = activate,
    .query_formats  = query_formats,
    .inputs         = NULL,
    .outputs        = avfilter_af_amix_outputs,
    .process_command = process_command,
    .flags          = AVFILTER_FLAG_DYNAMIC_INPUTS,
};


Generated by: GCOVR (Version 4.2)

Line	Branch	Exec	Source
1			/*
2			* Audio Mix Filter
3			* Copyright (c) 2012 Justin Ruggles <justin.ruggles@gmail.com>
4			*
5			* This file is part of FFmpeg.
6			*
7			* FFmpeg is free software; you can redistribute it and/or
8			* modify it under the terms of the GNU Lesser General Public
9			* License as published by the Free Software Foundation; either
10			* version 2.1 of the License, or (at your option) any later version.
11			*
12			* FFmpeg is distributed in the hope that it will be useful,
13			* but WITHOUT ANY WARRANTY; without even the implied warranty of
14			* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15			* Lesser General Public License for more details.
16			*
17			* You should have received a copy of the GNU Lesser General Public
18			* License along with FFmpeg; if not, write to the Free Software
19			* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20			*/
21
22			/**
23			* @file
24			* Audio Mix Filter
25			*
26			* Mixes audio from multiple sources into a single output. The channel layout,
27			* sample rate, and sample format will be the same for all inputs and the
28			* output.
29			*/
30
31			#include "libavutil/attributes.h"
32			#include "libavutil/audio_fifo.h"
33			#include "libavutil/avassert.h"
34			#include "libavutil/avstring.h"
35			#include "libavutil/channel_layout.h"
36			#include "libavutil/common.h"
37			#include "libavutil/eval.h"
38			#include "libavutil/float_dsp.h"
39			#include "libavutil/mathematics.h"
40			#include "libavutil/opt.h"
41			#include "libavutil/samplefmt.h"
42
43			#include "audio.h"
44			#include "avfilter.h"
45			#include "filters.h"
46			#include "formats.h"
47			#include "internal.h"
48
49			#define INPUT_ON 1 /*< input is active /
50			#define INPUT_EOF 2 /*< input has reached EOF (may still be active) /
51
52			#define DURATION_LONGEST 0
53			#define DURATION_SHORTEST 1
54			#define DURATION_FIRST 2
55
56
57			typedef struct FrameInfo {
58			int nb_samples;
59			int64_t pts;
60			struct FrameInfo *next;
61			} FrameInfo;
62
63			/**
64			* Linked list used to store timestamps and frame sizes of all frames in the
65			* FIFO for the first input.
66			*
67			* This is needed to keep timestamps synchronized for the case where multiple
68			* input frames are pushed to the filter for processing before a frame is
69			* requested by the output link.
70			*/
71			typedef struct FrameList {
72			int nb_frames;
73			int nb_samples;
74			FrameInfo *list;
75			FrameInfo *end;
76			} FrameList;
77
78		612	static void frame_list_clear(FrameList *frame_list)
79			{
80	✓✓	612	if (frame_list) {
81	✓✓	1214	while (frame_list->list) {
82		605	FrameInfo *info = frame_list->list;
83		605	frame_list->list = info->next;
84		605	av_free(info);
85			}
86		609	frame_list->nb_frames = 0;
87		609	frame_list->nb_samples = 0;
88		609	frame_list->end = NULL;
89			}
90		612	}
91
92		2046	static int frame_list_next_frame_size(FrameList *frame_list)
93			{
94	✓✓	2046	if (!frame_list->list)
95		1	return 0;
96		2045	return frame_list->list->nb_samples;
97			}
98
99		609	static int64_t frame_list_next_pts(FrameList *frame_list)
100			{
101	✓✓	609	if (!frame_list->list)
102		1	return AV_NOPTS_VALUE;
103		608	return frame_list->list->pts;
104			}
105
106		609	static void frame_list_remove_samples(FrameList *frame_list, int nb_samples)
107			{
108	✓✓	609	if (nb_samples >= frame_list->nb_samples) {
109		606	frame_list_clear(frame_list);
110			} else {
111		3	int samples = nb_samples;
112	✓✓	6	while (samples > 0) {
113		3	FrameInfo *info = frame_list->list;
114	✗✓	3	av_assert0(info);
115	✗✓	3	if (info->nb_samples <= samples) {
116			samples -= info->nb_samples;
117			frame_list->list = info->next;
118			if (!frame_list->list)
119			frame_list->end = NULL;
120			frame_list->nb_frames--;
121			frame_list->nb_samples -= info->nb_samples;
122			av_free(info);
123			} else {
124		3	info->nb_samples -= samples;
125		3	info->pts += samples;
126		3	frame_list->nb_samples -= samples;
127		3	samples = 0;
128			}
129			}
130			}
131		609	}
132
133		605	static int frame_list_add_frame(FrameList *frame_list, int nb_samples, int64_t pts)
134			{
135		605	FrameInfo info = av_malloc(sizeof(info));
136	✗✓	605	if (!info)
137			return AVERROR(ENOMEM);
138		605	info->nb_samples = nb_samples;
139		605	info->pts = pts;
140		605	info->next = NULL;
141
142	✓✗	605	if (!frame_list->list) {
143		605	frame_list->list = info;
144		605	frame_list->end = info;
145			} else {
146			av_assert0(frame_list->end);
147			frame_list->end->next = info;
148			frame_list->end = info;
149			}
150		605	frame_list->nb_frames++;
151		605	frame_list->nb_samples += nb_samples;
152
153		605	return 0;
154			}
155
156			/* FIXME: use directly links fifo */
157
158			typedef struct MixContext {
159			const AVClass class; /< class for AVOptions /
160			AVFloatDSPContext *fdsp;
161
162			int nb_inputs; /*< number of inputs /
163			int active_inputs; /*< number of input currently active /
164			int duration_mode; /*< mode for determining duration /
165			float dropout_transition; /*< transition time when an input drops out /
166			char weights_str; /< string for custom weights for every input /
167			int normalize; /*< if inputs are scaled /
168
169			int nb_channels; /*< number of channels /
170			int sample_rate; /*< sample rate /
171			int planar;
172			AVAudioFifo fifos; /< audio fifo for each input */
173			uint8_t input_state; /< current state of each input /
174			float input_scale; /< mixing scale factor for each input /
175			float weights; /< custom weights for every input /
176			float weight_sum; /*< sum of custom weights for every input /
177			float scale_norm; /< normalization factor for every input /
178			int64_t next_pts; /*< calculated pts for next output frame /
179			FrameList frame_list; /< list of frame info for the first input /
180			} MixContext;
181
182			#define OFFSET(x) offsetof(MixContext, x)
183			#define A AV_OPT_FLAG_AUDIO_PARAM
184			#define F AV_OPT_FLAG_FILTERING_PARAM
185			#define T AV_OPT_FLAG_RUNTIME_PARAM
186			static const AVOption amix_options[] = {
187			{ "inputs", "Number of inputs.",
188			OFFSET(nb_inputs), AV_OPT_TYPE_INT, { .i64 = 2 }, 1, INT16_MAX, A\|F },
189			{ "duration", "How to determine the end-of-stream.",
190			OFFSET(duration_mode), AV_OPT_TYPE_INT, { .i64 = DURATION_LONGEST }, 0, 2, A\|F, "duration" },
191			{ "longest", "Duration of longest input.", 0, AV_OPT_TYPE_CONST, { .i64 = DURATION_LONGEST }, 0, 0, A\|F, "duration" },
192			{ "shortest", "Duration of shortest input.", 0, AV_OPT_TYPE_CONST, { .i64 = DURATION_SHORTEST }, 0, 0, A\|F, "duration" },
193			{ "first", "Duration of first input.", 0, AV_OPT_TYPE_CONST, { .i64 = DURATION_FIRST }, 0, 0, A\|F, "duration" },
194			{ "dropout_transition", "Transition time, in seconds, for volume "
195			"renormalization when an input stream ends.",
196			OFFSET(dropout_transition), AV_OPT_TYPE_FLOAT, { .dbl = 2.0 }, 0, INT_MAX, A\|F },
197			{ "weights", "Set weight for each input.",
198			OFFSET(weights_str), AV_OPT_TYPE_STRING, {.str="1 1"}, 0, 0, A\|F\|T },
199			{ "normalize", "Scale inputs",
200			OFFSET(normalize), AV_OPT_TYPE_BOOL, {.i64=1}, 0, 1, A\|F\|T },
201			{ NULL }
202			};
203
204			AVFILTER_DEFINE_CLASS(amix);
205
206			/**
207			* Update the scaling factors to apply to each input during mixing.
208			*
209			* This balances the full volume range between active inputs and handles
210			* volume transitions when EOF is encountered on an input but mixing continues
211			* with the remaining inputs.
212			*/
213		612	static void calculate_scales(MixContext *s, int nb_samples)
214			{
215		612	float weight_sum = 0.f;
216			int i;
217
218	✓✓	2098	for (i = 0; i < s->nb_inputs; i++)
219	✓✓	1486	if (s->input_state[i] & INPUT_ON)
220		1095	weight_sum += FFABS(s->weights[i]);
221
222	✓✓	2098	for (i = 0; i < s->nb_inputs; i++) {
223	✓✓	1486	if (s->input_state[i] & INPUT_ON) {
224	✓✓	1095	if (s->scale_norm[i] > weight_sum / FFABS(s->weights[i])) {
225		153	s->scale_norm[i] -= ((s->weight_sum / FFABS(s->weights[i])) / s->nb_inputs) *
226		153	nb_samples / (s->dropout_transition * s->sample_rate);
227	✓✓	153	s->scale_norm[i] = FFMAX(s->scale_norm[i], weight_sum / FFABS(s->weights[i]));
228			}
229			}
230			}
231
232	✓✓	2098	for (i = 0; i < s->nb_inputs; i++) {
233	✓✓	1486	if (s->input_state[i] & INPUT_ON) {
234	✗✓	1095	if (!s->normalize)
235			s->input_scale[i] = FFABS(s->weights[i]);
236			else
237	✓✗	1095	s->input_scale[i] = 1.0f / s->scale_norm[i] * FFSIGN(s->weights[i]);
238			} else {
239		391	s->input_scale[i] = 0.0f;
240			}
241			}
242		612	}
243
244		3	static int config_output(AVFilterLink *outlink)
245			{
246		3	AVFilterContext *ctx = outlink->src;
247		3	MixContext *s = ctx->priv;
248			int i;
249			char buf[64];
250
251		3	s->planar = av_sample_fmt_is_planar(outlink->format);
252		3	s->sample_rate = outlink->sample_rate;
253		3	outlink->time_base = (AVRational){ 1, outlink->sample_rate };
254		3	s->next_pts = AV_NOPTS_VALUE;
255
256		3	s->frame_list = av_mallocz(sizeof(*s->frame_list));
257	✗✓	3	if (!s->frame_list)
258			return AVERROR(ENOMEM);
259
260		3	s->fifos = av_mallocz_array(s->nb_inputs, sizeof(*s->fifos));
261	✗✓	3	if (!s->fifos)
262			return AVERROR(ENOMEM);
263
264		3	s->nb_channels = outlink->channels;
265	✓✓	10	for (i = 0; i < s->nb_inputs; i++) {
266		7	s->fifos[i] = av_audio_fifo_alloc(outlink->format, s->nb_channels, 1024);
267	✗✓	7	if (!s->fifos[i])
268			return AVERROR(ENOMEM);
269			}
270
271		3	s->input_state = av_malloc(s->nb_inputs);
272	✗✓	3	if (!s->input_state)
273			return AVERROR(ENOMEM);
274		3	memset(s->input_state, INPUT_ON, s->nb_inputs);
275		3	s->active_inputs = s->nb_inputs;
276
277		3	s->input_scale = av_mallocz_array(s->nb_inputs, sizeof(*s->input_scale));
278		3	s->scale_norm = av_mallocz_array(s->nb_inputs, sizeof(*s->scale_norm));
279	✓✗✗✓	3	if (!s->input_scale \|\| !s->scale_norm)
280			return AVERROR(ENOMEM);
281	✓✓	10	for (i = 0; i < s->nb_inputs; i++)
282		7	s->scale_norm[i] = s->weight_sum / FFABS(s->weights[i]);
283		3	calculate_scales(s, 0);
284
285		3	av_get_channel_layout_string(buf, sizeof(buf), -1, outlink->channel_layout);
286
287		3	av_log(ctx, AV_LOG_VERBOSE,
288			"inputs:%d fmt:%s srate:%d cl:%s\n", s->nb_inputs,
289		3	av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf);
290
291		3	return 0;
292			}
293
294			/**
295			* Read samples from the input FIFOs, mix, and write to the output link.
296			*/
297		1564	static int output_frame(AVFilterLink *outlink)
298			{
299		1564	AVFilterContext *ctx = outlink->src;
300		1564	MixContext *s = ctx->priv;
301			AVFrame out_buf, in_buf;
302			int nb_samples, ns, i;
303
304	✓✗	1564	if (s->input_state[0] & INPUT_ON) {
305			/* first input live: use the corresponding frame size */
306		1564	nb_samples = frame_list_next_frame_size(s->frame_list);
307	✓✓	2609	for (i = 1; i < s->nb_inputs; i++) {
308	✓✓	2000	if (s->input_state[i] & INPUT_ON) {
309		1609	ns = av_audio_fifo_size(s->fifos[i]);
310	✓✓	1609	if (ns < nb_samples) {
311	✓✓	958	if (!(s->input_state[i] & INPUT_EOF))
312			/* unclosed input with not enough samples */
313		955	return 0;
314			/* closed input to drain */
315		3	nb_samples = ns;
316			}
317			}
318			}
319
320		609	s->next_pts = frame_list_next_pts(s->frame_list);
321			} else {
322			/* first input closed: use the available samples */
323			nb_samples = INT_MAX;
324			for (i = 1; i < s->nb_inputs; i++) {
325			if (s->input_state[i] & INPUT_ON) {
326			ns = av_audio_fifo_size(s->fifos[i]);
327			nb_samples = FFMIN(nb_samples, ns);
328			}
329			}
330			if (nb_samples == INT_MAX) {
331			ff_outlink_set_status(outlink, AVERROR_EOF, s->next_pts);
332			return 0;
333			}
334			}
335
336		609	frame_list_remove_samples(s->frame_list, nb_samples);
337
338		609	calculate_scales(s, nb_samples);
339
340	✓✓	609	if (nb_samples == 0)
341		1	return 0;
342
343		608	out_buf = ff_get_audio_buffer(outlink, nb_samples);
344	✗✓	608	if (!out_buf)
345			return AVERROR(ENOMEM);
346
347		608	in_buf = ff_get_audio_buffer(outlink, nb_samples);
348	✗✓	608	if (!in_buf) {
349			av_frame_free(&out_buf);
350			return AVERROR(ENOMEM);
351			}
352
353	✓✓	2085	for (i = 0; i < s->nb_inputs; i++) {
354	✓✓	1477	if (s->input_state[i] & INPUT_ON) {
355			int planes, plane_size, p;
356
357		1086	av_audio_fifo_read(s->fifos[i], (void **)in_buf->extended_data,
358			nb_samples);
359
360	✗✓	1086	planes = s->planar ? s->nb_channels : 1;
361	✓✗	1086	plane_size = nb_samples * (s->planar ? 1 : s->nb_channels);
362		1086	plane_size = FFALIGN(plane_size, 16);
363
364	✗✓	1086	if (out_buf->format == AV_SAMPLE_FMT_FLT \|\|
365			out_buf->format == AV_SAMPLE_FMT_FLTP) {
366	✓✓	2172	for (p = 0; p < planes; p++) {
367		1086	s->fdsp->vector_fmac_scalar((float *)out_buf->extended_data[p],
368		1086	(float *) in_buf->extended_data[p],
369		1086	s->input_scale[i], plane_size);
370			}
371			} else {
372			for (p = 0; p < planes; p++) {
373			s->fdsp->vector_dmac_scalar((double *)out_buf->extended_data[p],
374			(double *) in_buf->extended_data[p],
375			s->input_scale[i], plane_size);
376			}
377			}
378			}
379			}
380		608	av_frame_free(&in_buf);
381
382		608	out_buf->pts = s->next_pts;
383	✓✗	608	if (s->next_pts != AV_NOPTS_VALUE)
384		608	s->next_pts += nb_samples;
385
386		608	return ff_filter_frame(outlink, out_buf);
387			}
388
389			/**
390			* Requests a frame, if needed, from each input link other than the first.
391			*/
392		482	static int request_samples(AVFilterContext *ctx, int min_samples)
393			{
394		482	MixContext *s = ctx->priv;
395			int i;
396
397	✗✓	482	av_assert0(s->nb_inputs > 1);
398
399	✓✓	1228	for (i = 1; i < s->nb_inputs; i++) {
400	✓✓	746	if (!(s->input_state[i] & INPUT_ON) \|\|
401	✓✓	654	(s->input_state[i] & INPUT_EOF))
402		95	continue;
403	✓✓	651	if (av_audio_fifo_size(s->fifos[i]) >= min_samples)
404		89	continue;
405		562	ff_inlink_request_frame(ctx->inputs[i]);
406			}
407		482	return output_frame(ctx->outputs[0]);
408			}
409
410			/**
411			* Calculates the number of active inputs and determines EOF based on the
412			* duration option.
413			*
414			* @return 0 if mixing should continue, or AVERROR_EOF if mixing should stop.
415			*/
416		1696	static int calc_active_inputs(MixContext *s)
417			{
418			int i;
419		1696	int active_inputs = 0;
420	✓✓	5871	for (i = 0; i < s->nb_inputs; i++)
421		4175	active_inputs += !!(s->input_state[i] & INPUT_ON);
422		1696	s->active_inputs = active_inputs;
423
424	✓✓	1696	if (!active_inputs \|\|
425	✓✓✓✓	1694	(s->duration_mode == DURATION_FIRST && !(s->input_state[0] & INPUT_ON)) \|\|
426	✗✓✗✗	1693	(s->duration_mode == DURATION_SHORTEST && active_inputs != s->nb_inputs))
427		3	return AVERROR_EOF;
428		1693	return 0;
429			}
430
431		1699	static int activate(AVFilterContext *ctx)
432			{
433		1699	AVFilterLink *outlink = ctx->outputs[0];
434		1699	MixContext *s = ctx->priv;
435		1699	AVFrame *buf = NULL;
436			int i, ret;
437
438	✓✓✓✓	1706	FF_FILTER_FORWARD_STATUS_BACK_ALL(outlink, ctx);
439
440	✓✓	5871	for (i = 0; i < s->nb_inputs; i++) {
441		4175	AVFilterLink *inlink = ctx->inputs[i];
442
443	✓✓	4175	if ((ret = ff_inlink_consume_frame(ctx->inputs[i], &buf)) > 0) {
444	✓✓	1082	if (i == 0) {
445		605	int64_t pts = av_rescale_q(buf->pts, inlink->time_base,
446			outlink->time_base);
447		605	ret = frame_list_add_frame(s->frame_list, buf->nb_samples, pts);
448	✗✓	605	if (ret < 0) {
449			av_frame_free(&buf);
450			return ret;
451			}
452			}
453
454		1082	ret = av_audio_fifo_write(s->fifos[i], (void **)buf->extended_data,
455		1082	buf->nb_samples);
456	✗✓	1082	if (ret < 0) {
457			av_frame_free(&buf);
458			return ret;
459			}
460
461		1082	av_frame_free(&buf);
462
463		1082	ret = output_frame(outlink);
464	✗✓	1082	if (ret < 0)
465			return ret;
466			}
467			}
468
469	✓✓	5871	for (i = 0; i < s->nb_inputs; i++) {
470			int64_t pts;
471			int status;
472
473	✓✓	4175	if (ff_inlink_acknowledge_status(ctx->inputs[i], &status, &pts)) {
474	✓✗	877	if (status == AVERROR_EOF) {
475	✓✓	877	if (i == 0) {
476		3	s->input_state[i] = 0;
477	✗✓	3	if (s->nb_inputs == 1) {
478			ff_outlink_set_status(outlink, status, pts);
479			return 0;
480			}
481			} else {
482		874	s->input_state[i] \|= INPUT_EOF;
483	✓✓	874	if (av_audio_fifo_size(s->fifos[i]) == 0) {
484		871	s->input_state[i] = 0;
485			}
486			}
487			}
488			}
489			}
490
491	✓✓	1696	if (calc_active_inputs(s)) {
492		3	ff_outlink_set_status(outlink, AVERROR_EOF, s->next_pts);
493		3	return 0;
494			}
495
496	✓✓	1693	if (ff_outlink_frame_wanted(outlink)) {
497			int wanted_samples;
498
499	✗✓	1087	if (!(s->input_state[0] & INPUT_ON))
500			return request_samples(ctx, 1);
501
502	✓✓	1087	if (s->frame_list->nb_frames == 0) {
503		605	ff_inlink_request_frame(ctx->inputs[0]);
504		605	return 0;
505			}
506	✗✓	482	av_assert0(s->frame_list->nb_frames > 0);
507
508		482	wanted_samples = frame_list_next_frame_size(s->frame_list);
509
510		482	return request_samples(ctx, wanted_samples);
511			}
512
513		606	return 0;
514			}
515
516		6	static void parse_weights(AVFilterContext *ctx)
517			{
518		6	MixContext *s = ctx->priv;
519		6	float last_weight = 1.f;
520			char *p;
521			int i;
522
523		6	s->weight_sum = 0.f;
524		6	p = s->weights_str;
525	✓✗	12	for (i = 0; i < s->nb_inputs; i++) {
526		12	last_weight = av_strtod(p, &p);
527		12	s->weights[i] = last_weight;
528		12	s->weight_sum += FFABS(last_weight);
529	✓✗✓✓	12	if (p && *p) {
530		6	p++;
531			} else {
532		6	i++;
533		6	break;
534			}
535			}
536
537	✓✓	8	for (; i < s->nb_inputs; i++) {
538		2	s->weights[i] = last_weight;
539		2	s->weight_sum += FFABS(last_weight);
540			}
541		6	}
542
543		6	static av_cold int init(AVFilterContext *ctx)
544			{
545		6	MixContext *s = ctx->priv;
546			int i, ret;
547
548	✓✓	20	for (i = 0; i < s->nb_inputs; i++) {
549		14	AVFilterPad pad = { 0 };
550
551		14	pad.type = AVMEDIA_TYPE_AUDIO;
552		14	pad.name = av_asprintf("input%d", i);
553	✗✓	14	if (!pad.name)
554			return AVERROR(ENOMEM);
555
556	✗✓	14	if ((ret = ff_insert_inpad(ctx, i, &pad)) < 0) {
557			av_freep(&pad.name);
558			return ret;
559			}
560			}
561
562		6	s->fdsp = avpriv_float_dsp_alloc(0);
563	✗✓	6	if (!s->fdsp)
564			return AVERROR(ENOMEM);
565
566		6	s->weights = av_mallocz_array(s->nb_inputs, sizeof(*s->weights));
567	✗✓	6	if (!s->weights)
568			return AVERROR(ENOMEM);
569
570		6	parse_weights(ctx);
571
572		6	return 0;
573			}
574
575		6	static av_cold void uninit(AVFilterContext *ctx)
576			{
577			int i;
578		6	MixContext *s = ctx->priv;
579
580	✓✓	6	if (s->fifos) {
581	✓✓	10	for (i = 0; i < s->nb_inputs; i++)
582		7	av_audio_fifo_free(s->fifos[i]);
583		3	av_freep(&s->fifos);
584			}
585		6	frame_list_clear(s->frame_list);
586		6	av_freep(&s->frame_list);
587		6	av_freep(&s->input_state);
588		6	av_freep(&s->input_scale);
589		6	av_freep(&s->scale_norm);
590		6	av_freep(&s->weights);
591		6	av_freep(&s->fdsp);
592
593	✓✓	20	for (i = 0; i < ctx->nb_inputs; i++)
594		14	av_freep(&ctx->input_pads[i].name);
595		6	}
596
597		3	static int query_formats(AVFilterContext *ctx)
598			{
599			static const enum AVSampleFormat sample_fmts[] = {
600			AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
601			AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBLP,
602			AV_SAMPLE_FMT_NONE
603			};
604			int ret;
605
606	✓✗✗✓	6	if ((ret = ff_set_common_formats(ctx, ff_make_format_list(sample_fmts))) < 0 \|\|
607		3	(ret = ff_set_common_samplerates(ctx, ff_all_samplerates())) < 0)
608			return ret;
609
610		3	return ff_set_common_channel_layouts(ctx, ff_all_channel_counts());
611			}
612
613			static int process_command(AVFilterContext ctx, const char cmd, const char *args,
614			char *res, int res_len, int flags)
615			{
616			MixContext *s = ctx->priv;
617			int ret;
618
619			ret = ff_filter_process_command(ctx, cmd, args, res, res_len, flags);
620			if (ret < 0)
621			return ret;
622
623			parse_weights(ctx);
624			for (int i = 0; i < s->nb_inputs; i++)
625			s->scale_norm[i] = s->weight_sum / FFABS(s->weights[i]);
626			calculate_scales(s, 0);
627
628			return 0;
629			}
630
631			static const AVFilterPad avfilter_af_amix_outputs[] = {
632			{
633			.name = "default",
634			.type = AVMEDIA_TYPE_AUDIO,
635			.config_props = config_output,
636			},
637			{ NULL }
638			};
639
640			AVFilter ff_af_amix = {
641			.name = "amix",
642			.description = NULL_IF_CONFIG_SMALL("Audio mixing."),
643			.priv_size = sizeof(MixContext),
644			.priv_class = &amix_class,
645			.init = init,
646			.uninit = uninit,
647			.activate = activate,
648			.query_formats = query_formats,
649			.inputs = NULL,
650			.outputs = avfilter_af_amix_outputs,
651			.process_command = process_command,
652			.flags = AVFILTER_FLAG_DYNAMIC_INPUTS,
653			};