FFmpeg coverage

GCC Code Coverage Report

Directory:	../../../ffmpeg/		Exec	Total	Coverage
File:	src/libavfilter/f_select.c	Lines:	166	202	82.2 %
Date:	2021-01-22 05:18:52	Branches:	64	108	59.3 %


/*
 * Copyright (c) 2011 Stefano Sabatini
 *
 * 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
 * filter for selecting which frame passes in the filterchain
 */

#include "libavutil/avstring.h"
#include "libavutil/eval.h"
#include "libavutil/fifo.h"
#include "libavutil/imgutils.h"
#include "libavutil/internal.h"
#include "libavutil/opt.h"
#include "libavutil/pixdesc.h"
#include "avfilter.h"
#include "audio.h"
#include "formats.h"
#include "internal.h"
#include "video.h"
#include "scene_sad.h"

static const char *const var_names[] = {
    "TB",                ///< timebase

    "pts",               ///< original pts in the file of the frame
    "start_pts",         ///< first PTS in the stream, expressed in TB units
    "prev_pts",          ///< previous frame PTS
    "prev_selected_pts", ///< previous selected frame PTS

    "t",                 ///< timestamp expressed in seconds
    "start_t",           ///< first PTS in the stream, expressed in seconds
    "prev_t",            ///< previous frame time
    "prev_selected_t",   ///< previously selected time

    "pict_type",         ///< the type of picture in the movie
    "I",
    "P",
    "B",
    "S",
    "SI",
    "SP",
    "BI",
    "PICT_TYPE_I",
    "PICT_TYPE_P",
    "PICT_TYPE_B",
    "PICT_TYPE_S",
    "PICT_TYPE_SI",
    "PICT_TYPE_SP",
    "PICT_TYPE_BI",

    "interlace_type",    ///< the frame interlace type
    "PROGRESSIVE",
    "TOPFIRST",
    "BOTTOMFIRST",

    "consumed_samples_n",///< number of samples consumed by the filter (only audio)
    "samples_n",         ///< number of samples in the current frame (only audio)
    "sample_rate",       ///< sample rate (only audio)

    "n",                 ///< frame number (starting from zero)
    "selected_n",        ///< selected frame number (starting from zero)
    "prev_selected_n",   ///< number of the last selected frame

    "key",               ///< tell if the frame is a key frame
    "pos",               ///< original position in the file of the frame

    "scene",

    "concatdec_select",  ///< frame is within the interval set by the concat demuxer

    NULL
};

enum var_name {
    VAR_TB,

    VAR_PTS,
    VAR_START_PTS,
    VAR_PREV_PTS,
    VAR_PREV_SELECTED_PTS,

    VAR_T,
    VAR_START_T,
    VAR_PREV_T,
    VAR_PREV_SELECTED_T,

    VAR_PICT_TYPE,
    VAR_I,
    VAR_P,
    VAR_B,
    VAR_S,
    VAR_SI,
    VAR_SP,
    VAR_BI,
    VAR_PICT_TYPE_I,
    VAR_PICT_TYPE_P,
    VAR_PICT_TYPE_B,
    VAR_PICT_TYPE_S,
    VAR_PICT_TYPE_SI,
    VAR_PICT_TYPE_SP,
    VAR_PICT_TYPE_BI,

    VAR_INTERLACE_TYPE,
    VAR_INTERLACE_TYPE_P,
    VAR_INTERLACE_TYPE_T,
    VAR_INTERLACE_TYPE_B,

    VAR_CONSUMED_SAMPLES_N,
    VAR_SAMPLES_N,
    VAR_SAMPLE_RATE,

    VAR_N,
    VAR_SELECTED_N,
    VAR_PREV_SELECTED_N,

    VAR_KEY,
    VAR_POS,

    VAR_SCENE,

    VAR_CONCATDEC_SELECT,

    VAR_VARS_NB
};

typedef struct SelectContext {
    const AVClass *class;
    char *expr_str;
    AVExpr *expr;
    double var_values[VAR_VARS_NB];
    int bitdepth;
    int nb_planes;
    ptrdiff_t width[4];
    ptrdiff_t height[4];
    int do_scene_detect;            ///< 1 if the expression requires scene detection variables, 0 otherwise
    ff_scene_sad_fn sad;            ///< Sum of the absolute difference function (scene detect only)
    double prev_mafd;               ///< previous MAFD                           (scene detect only)
    AVFrame *prev_picref;           ///< previous frame                          (scene detect only)
    double select;
    int select_out;                 ///< mark the selected output pad index
    int nb_outputs;
} SelectContext;

#define OFFSET(x) offsetof(SelectContext, x)
#define DEFINE_OPTIONS(filt_name, FLAGS)                            \
static const AVOption filt_name##_options[] = {                     \
    { "expr", "set an expression to use for selecting frames", OFFSET(expr_str), AV_OPT_TYPE_STRING, { .str = "1" }, .flags=FLAGS }, \
    { "e",    "set an expression to use for selecting frames", OFFSET(expr_str), AV_OPT_TYPE_STRING, { .str = "1" }, .flags=FLAGS }, \
    { "outputs", "set the number of outputs", OFFSET(nb_outputs), AV_OPT_TYPE_INT, {.i64 = 1}, 1, INT_MAX, .flags=FLAGS }, \
    { "n",       "set the number of outputs", OFFSET(nb_outputs), AV_OPT_TYPE_INT, {.i64 = 1}, 1, INT_MAX, .flags=FLAGS }, \
    { NULL }                                                            \
}

static int request_frame(AVFilterLink *outlink);

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

    if ((ret = av_expr_parse(&select->expr, select->expr_str,
                             var_names, NULL, NULL, NULL, NULL, 0, ctx)) < 0) {
        av_log(ctx, AV_LOG_ERROR, "Error while parsing expression '%s'\n",
               select->expr_str);
        return ret;
    }
    select->do_scene_detect = !!strstr(select->expr_str, "scene");

    for (i = 0; i < select->nb_outputs; i++) {
        AVFilterPad pad = { 0 };

        pad.name = av_asprintf("output%d", i);
        if (!pad.name)
            return AVERROR(ENOMEM);
        pad.type = ctx->filter->inputs[0].type;
        pad.request_frame = request_frame;
        if ((ret = ff_insert_outpad(ctx, i, &pad)) < 0) {
            av_freep(&pad.name);
            return ret;
        }
    }

    return 0;
}

#define INTERLACE_TYPE_P 0
#define INTERLACE_TYPE_T 1
#define INTERLACE_TYPE_B 2

static int config_input(AVFilterLink *inlink)
{
    SelectContext *select = inlink->dst->priv;
    const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
    int is_yuv = !(desc->flags & AV_PIX_FMT_FLAG_RGB) &&
                 (desc->flags & AV_PIX_FMT_FLAG_PLANAR) &&
                 desc->nb_components >= 3;

    select->bitdepth = desc->comp[0].depth;
    select->nb_planes = is_yuv ? 1 : av_pix_fmt_count_planes(inlink->format);

    for (int plane = 0; plane < select->nb_planes; plane++) {
        ptrdiff_t line_size = av_image_get_linesize(inlink->format, inlink->w, plane);
        int vsub = desc->log2_chroma_h;

        select->width[plane] = line_size >> (select->bitdepth > 8);
        select->height[plane] = plane == 1 || plane == 2 ?  AV_CEIL_RSHIFT(inlink->h, vsub) : inlink->h;
    }

    select->var_values[VAR_N]          = 0.0;
    select->var_values[VAR_SELECTED_N] = 0.0;

    select->var_values[VAR_TB] = av_q2d(inlink->time_base);

    select->var_values[VAR_PREV_PTS]          = NAN;
    select->var_values[VAR_PREV_SELECTED_PTS] = NAN;
    select->var_values[VAR_PREV_SELECTED_T]   = NAN;
    select->var_values[VAR_PREV_T]            = NAN;
    select->var_values[VAR_START_PTS]         = NAN;
    select->var_values[VAR_START_T]           = NAN;

    select->var_values[VAR_I]  = AV_PICTURE_TYPE_I;
    select->var_values[VAR_P]  = AV_PICTURE_TYPE_P;
    select->var_values[VAR_B]  = AV_PICTURE_TYPE_B;
    select->var_values[VAR_SI] = AV_PICTURE_TYPE_SI;
    select->var_values[VAR_SP] = AV_PICTURE_TYPE_SP;
    select->var_values[VAR_BI] = AV_PICTURE_TYPE_BI;
    select->var_values[VAR_PICT_TYPE_I]  = AV_PICTURE_TYPE_I;
    select->var_values[VAR_PICT_TYPE_P]  = AV_PICTURE_TYPE_P;
    select->var_values[VAR_PICT_TYPE_B]  = AV_PICTURE_TYPE_B;
    select->var_values[VAR_PICT_TYPE_SI] = AV_PICTURE_TYPE_SI;
    select->var_values[VAR_PICT_TYPE_SP] = AV_PICTURE_TYPE_SP;
    select->var_values[VAR_PICT_TYPE_BI] = AV_PICTURE_TYPE_BI;

    select->var_values[VAR_INTERLACE_TYPE_P] = INTERLACE_TYPE_P;
    select->var_values[VAR_INTERLACE_TYPE_T] = INTERLACE_TYPE_T;
    select->var_values[VAR_INTERLACE_TYPE_B] = INTERLACE_TYPE_B;

    select->var_values[VAR_PICT_TYPE]         = NAN;
    select->var_values[VAR_INTERLACE_TYPE]    = NAN;
    select->var_values[VAR_SCENE]             = NAN;
    select->var_values[VAR_CONSUMED_SAMPLES_N] = NAN;
    select->var_values[VAR_SAMPLES_N]          = NAN;

    select->var_values[VAR_SAMPLE_RATE] =
        inlink->type == AVMEDIA_TYPE_AUDIO ? inlink->sample_rate : NAN;

    if (CONFIG_SELECT_FILTER && select->do_scene_detect) {
        select->sad = ff_scene_sad_get_fn(select->bitdepth == 8 ? 8 : 16);
        if (!select->sad)
            return AVERROR(EINVAL);
    }
    return 0;
}

static double get_scene_score(AVFilterContext *ctx, AVFrame *frame)
{
    double ret = 0;
    SelectContext *select = ctx->priv;
    AVFrame *prev_picref = select->prev_picref;

    if (prev_picref &&
        frame->height == prev_picref->height &&
        frame->width  == prev_picref->width) {
        uint64_t sad = 0;
        double mafd, diff;
        uint64_t count = 0;

        for (int plane = 0; plane < select->nb_planes; plane++) {
            uint64_t plane_sad;
            select->sad(prev_picref->data[plane], prev_picref->linesize[plane],
                    frame->data[plane], frame->linesize[plane],
                    select->width[plane], select->height[plane], &plane_sad);
            sad += plane_sad;
            count += select->width[plane] * select->height[plane];
        }

        emms_c();
        mafd = (double)sad / count / (1ULL << (select->bitdepth - 8));
        diff = fabs(mafd - select->prev_mafd);
        ret  = av_clipf(FFMIN(mafd, diff) / 100., 0, 1);
        select->prev_mafd = mafd;
        av_frame_free(&prev_picref);
    }
    select->prev_picref = av_frame_clone(frame);
    return ret;
}

static double get_concatdec_select(AVFrame *frame, int64_t pts)
{
    AVDictionary *metadata = frame->metadata;
    AVDictionaryEntry *start_time_entry = av_dict_get(metadata, "lavf.concatdec.start_time", NULL, 0);
    AVDictionaryEntry *duration_entry = av_dict_get(metadata, "lavf.concatdec.duration", NULL, 0);
    if (start_time_entry) {
        int64_t start_time = strtoll(start_time_entry->value, NULL, 10);
        if (pts >= start_time) {
            if (duration_entry) {
              int64_t duration = strtoll(duration_entry->value, NULL, 10);
              if (pts < start_time + duration)
                  return -1;
              else
                  return 0;
            }
            return -1;
        }
        return 0;
    }
    return NAN;
}

static void select_frame(AVFilterContext *ctx, AVFrame *frame)
{
    SelectContext *select = ctx->priv;
    AVFilterLink *inlink = ctx->inputs[0];
    double res;

    if (isnan(select->var_values[VAR_START_PTS]))
        select->var_values[VAR_START_PTS] = TS2D(frame->pts);
    if (isnan(select->var_values[VAR_START_T]))
        select->var_values[VAR_START_T] = TS2D(frame->pts) * av_q2d(inlink->time_base);

    select->var_values[VAR_N  ] = inlink->frame_count_out;
    select->var_values[VAR_PTS] = TS2D(frame->pts);
    select->var_values[VAR_T  ] = TS2D(frame->pts) * av_q2d(inlink->time_base);
    select->var_values[VAR_POS] = frame->pkt_pos == -1 ? NAN : frame->pkt_pos;
    select->var_values[VAR_KEY] = frame->key_frame;
    select->var_values[VAR_CONCATDEC_SELECT] = get_concatdec_select(frame, av_rescale_q(frame->pts, inlink->time_base, AV_TIME_BASE_Q));

    switch (inlink->type) {
    case AVMEDIA_TYPE_AUDIO:
        select->var_values[VAR_SAMPLES_N] = frame->nb_samples;
        break;

    case AVMEDIA_TYPE_VIDEO:
        select->var_values[VAR_INTERLACE_TYPE] =
            !frame->interlaced_frame ? INTERLACE_TYPE_P :
        frame->top_field_first ? INTERLACE_TYPE_T : INTERLACE_TYPE_B;
        select->var_values[VAR_PICT_TYPE] = frame->pict_type;
        if (select->do_scene_detect) {
            char buf[32];
            select->var_values[VAR_SCENE] = get_scene_score(ctx, frame);
            // TODO: document metadata
            snprintf(buf, sizeof(buf), "%f", select->var_values[VAR_SCENE]);
            av_dict_set(&frame->metadata, "lavfi.scene_score", buf, 0);
        }
        break;
    }

    select->select = res = av_expr_eval(select->expr, select->var_values, NULL);
    av_log(inlink->dst, AV_LOG_DEBUG,
           "n:%f pts:%f t:%f key:%d",
           select->var_values[VAR_N],
           select->var_values[VAR_PTS],
           select->var_values[VAR_T],
           frame->key_frame);

    switch (inlink->type) {
    case AVMEDIA_TYPE_VIDEO:
        av_log(inlink->dst, AV_LOG_DEBUG, " interlace_type:%c pict_type:%c scene:%f",
               (!frame->interlaced_frame) ? 'P' :
               frame->top_field_first     ? 'T' : 'B',
               av_get_picture_type_char(frame->pict_type),
               select->var_values[VAR_SCENE]);
        break;
    case AVMEDIA_TYPE_AUDIO:
        av_log(inlink->dst, AV_LOG_DEBUG, " samples_n:%d consumed_samples_n:%f",
               frame->nb_samples,
               select->var_values[VAR_CONSUMED_SAMPLES_N]);
        break;
    }

    if (res == 0) {
        select->select_out = -1; /* drop */
    } else if (isnan(res) || res < 0) {
        select->select_out = 0; /* first output */
    } else {
        select->select_out = FFMIN(ceilf(res)-1, select->nb_outputs-1); /* other outputs */
    }

    av_log(inlink->dst, AV_LOG_DEBUG, " -> select:%f select_out:%d\n", res, select->select_out);

    if (res) {
        select->var_values[VAR_PREV_SELECTED_N]   = select->var_values[VAR_N];
        select->var_values[VAR_PREV_SELECTED_PTS] = select->var_values[VAR_PTS];
        select->var_values[VAR_PREV_SELECTED_T]   = select->var_values[VAR_T];
        select->var_values[VAR_SELECTED_N] += 1.0;
        if (inlink->type == AVMEDIA_TYPE_AUDIO)
            select->var_values[VAR_CONSUMED_SAMPLES_N] += frame->nb_samples;
    }

    select->var_values[VAR_PREV_PTS] = select->var_values[VAR_PTS];
    select->var_values[VAR_PREV_T]   = select->var_values[VAR_T];
}

static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
{
    AVFilterContext *ctx = inlink->dst;
    SelectContext *select = ctx->priv;

    select_frame(ctx, frame);
    if (select->select)
        return ff_filter_frame(ctx->outputs[select->select_out], frame);

    av_frame_free(&frame);
    return 0;
}

static int request_frame(AVFilterLink *outlink)
{
    AVFilterLink *inlink = outlink->src->inputs[0];
    int ret = ff_request_frame(inlink);
    return ret;
}

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

    av_expr_free(select->expr);
    select->expr = NULL;

    for (i = 0; i < ctx->nb_outputs; i++)
        av_freep(&ctx->output_pads[i].name);

    if (select->do_scene_detect) {
        av_frame_free(&select->prev_picref);
    }
}

#if CONFIG_ASELECT_FILTER

DEFINE_OPTIONS(aselect, AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM);
AVFILTER_DEFINE_CLASS(aselect);

static av_cold int aselect_init(AVFilterContext *ctx)
{
    SelectContext *select = ctx->priv;
    int ret;

    if ((ret = init(ctx)) < 0)
        return ret;

    if (select->do_scene_detect) {
        av_log(ctx, AV_LOG_ERROR, "Scene detection is ignored in aselect filter\n");
        return AVERROR(EINVAL);
    }

    return 0;
}

static const AVFilterPad avfilter_af_aselect_inputs[] = {
    {
        .name         = "default",
        .type         = AVMEDIA_TYPE_AUDIO,
        .config_props = config_input,
        .filter_frame = filter_frame,
    },
    { NULL }
};

AVFilter ff_af_aselect = {
    .name        = "aselect",
    .description = NULL_IF_CONFIG_SMALL("Select audio frames to pass in output."),
    .init        = aselect_init,
    .uninit      = uninit,
    .priv_size   = sizeof(SelectContext),
    .inputs      = avfilter_af_aselect_inputs,
    .priv_class  = &aselect_class,
    .flags       = AVFILTER_FLAG_DYNAMIC_OUTPUTS,
};
#endif /* CONFIG_ASELECT_FILTER */

#if CONFIG_SELECT_FILTER

static int query_formats(AVFilterContext *ctx)
{
    SelectContext *select = ctx->priv;

    if (!select->do_scene_detect) {
        return ff_default_query_formats(ctx);
    } else {
        int ret;
        static const enum AVPixelFormat pix_fmts[] = {
            AV_PIX_FMT_RGB24, AV_PIX_FMT_BGR24, AV_PIX_FMT_RGBA,
            AV_PIX_FMT_ABGR, AV_PIX_FMT_BGRA, AV_PIX_FMT_GRAY8,
            AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUVJ420P,
            AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUVJ422P,
            AV_PIX_FMT_YUV420P10,
            AV_PIX_FMT_NONE
        };
        AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);

        if (!fmts_list)
            return AVERROR(ENOMEM);
        ret = ff_set_common_formats(ctx, fmts_list);
        if (ret < 0)
            return ret;
    }
    return 0;
}

DEFINE_OPTIONS(select, AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM);
AVFILTER_DEFINE_CLASS(select);

static av_cold int select_init(AVFilterContext *ctx)
{
    int ret;

    if ((ret = init(ctx)) < 0)
        return ret;

    return 0;
}

static const AVFilterPad avfilter_vf_select_inputs[] = {
    {
        .name         = "default",
        .type         = AVMEDIA_TYPE_VIDEO,
        .config_props = config_input,
        .filter_frame = filter_frame,
    },
    { NULL }
};

AVFilter ff_vf_select = {
    .name          = "select",
    .description   = NULL_IF_CONFIG_SMALL("Select video frames to pass in output."),
    .init          = select_init,
    .uninit        = uninit,
    .query_formats = query_formats,
    .priv_size     = sizeof(SelectContext),
    .priv_class    = &select_class,
    .inputs        = avfilter_vf_select_inputs,
    .flags         = AVFILTER_FLAG_DYNAMIC_OUTPUTS,
};
#endif /* CONFIG_SELECT_FILTER */


Generated by: GCOVR (Version 4.2)

Line	Branch	Exec	Source
1			/*
2			* Copyright (c) 2011 Stefano Sabatini
3			*
4			* This file is part of FFmpeg.
5			*
6			* FFmpeg is free software; you can redistribute it and/or
7			* modify it under the terms of the GNU Lesser General Public
8			* License as published by the Free Software Foundation; either
9			* version 2.1 of the License, or (at your option) any later version.
10			*
11			* FFmpeg is distributed in the hope that it will be useful,
12			* but WITHOUT ANY WARRANTY; without even the implied warranty of
13			* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14			* Lesser General Public License for more details.
15			*
16			* You should have received a copy of the GNU Lesser General Public
17			* License along with FFmpeg; if not, write to the Free Software
18			* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19			*/
20
21			/**
22			* @file
23			* filter for selecting which frame passes in the filterchain
24			*/
25
26			#include "libavutil/avstring.h"
27			#include "libavutil/eval.h"
28			#include "libavutil/fifo.h"
29			#include "libavutil/imgutils.h"
30			#include "libavutil/internal.h"
31			#include "libavutil/opt.h"
32			#include "libavutil/pixdesc.h"
33			#include "avfilter.h"
34			#include "audio.h"
35			#include "formats.h"
36			#include "internal.h"
37			#include "video.h"
38			#include "scene_sad.h"
39
40			static const char *const var_names[] = {
41			"TB", ///< timebase
42
43			"pts", ///< original pts in the file of the frame
44			"start_pts", ///< first PTS in the stream, expressed in TB units
45			"prev_pts", ///< previous frame PTS
46			"prev_selected_pts", ///< previous selected frame PTS
47
48			"t", ///< timestamp expressed in seconds
49			"start_t", ///< first PTS in the stream, expressed in seconds
50			"prev_t", ///< previous frame time
51			"prev_selected_t", ///< previously selected time
52
53			"pict_type", ///< the type of picture in the movie
54			"I",
55			"P",
56			"B",
57			"S",
58			"SI",
59			"SP",
60			"BI",
61			"PICT_TYPE_I",
62			"PICT_TYPE_P",
63			"PICT_TYPE_B",
64			"PICT_TYPE_S",
65			"PICT_TYPE_SI",
66			"PICT_TYPE_SP",
67			"PICT_TYPE_BI",
68
69			"interlace_type", ///< the frame interlace type
70			"PROGRESSIVE",
71			"TOPFIRST",
72			"BOTTOMFIRST",
73
74			"consumed_samples_n",///< number of samples consumed by the filter (only audio)
75			"samples_n", ///< number of samples in the current frame (only audio)
76			"sample_rate", ///< sample rate (only audio)
77
78			"n", ///< frame number (starting from zero)
79			"selected_n", ///< selected frame number (starting from zero)
80			"prev_selected_n", ///< number of the last selected frame
81
82			"key", ///< tell if the frame is a key frame
83			"pos", ///< original position in the file of the frame
84
85			"scene",
86
87			"concatdec_select", ///< frame is within the interval set by the concat demuxer
88
89			NULL
90			};
91
92			enum var_name {
93			VAR_TB,
94
95			VAR_PTS,
96			VAR_START_PTS,
97			VAR_PREV_PTS,
98			VAR_PREV_SELECTED_PTS,
99
100			VAR_T,
101			VAR_START_T,
102			VAR_PREV_T,
103			VAR_PREV_SELECTED_T,
104
105			VAR_PICT_TYPE,
106			VAR_I,
107			VAR_P,
108			VAR_B,
109			VAR_S,
110			VAR_SI,
111			VAR_SP,
112			VAR_BI,
113			VAR_PICT_TYPE_I,
114			VAR_PICT_TYPE_P,
115			VAR_PICT_TYPE_B,
116			VAR_PICT_TYPE_S,
117			VAR_PICT_TYPE_SI,
118			VAR_PICT_TYPE_SP,
119			VAR_PICT_TYPE_BI,
120
121			VAR_INTERLACE_TYPE,
122			VAR_INTERLACE_TYPE_P,
123			VAR_INTERLACE_TYPE_T,
124			VAR_INTERLACE_TYPE_B,
125
126			VAR_CONSUMED_SAMPLES_N,
127			VAR_SAMPLES_N,
128			VAR_SAMPLE_RATE,
129
130			VAR_N,
131			VAR_SELECTED_N,
132			VAR_PREV_SELECTED_N,
133
134			VAR_KEY,
135			VAR_POS,
136
137			VAR_SCENE,
138
139			VAR_CONCATDEC_SELECT,
140
141			VAR_VARS_NB
142			};
143
144			typedef struct SelectContext {
145			const AVClass *class;
146			char *expr_str;
147			AVExpr *expr;
148			double var_values[VAR_VARS_NB];
149			int bitdepth;
150			int nb_planes;
151			ptrdiff_t width[4];
152			ptrdiff_t height[4];
153			int do_scene_detect; ///< 1 if the expression requires scene detection variables, 0 otherwise
154			ff_scene_sad_fn sad; ///< Sum of the absolute difference function (scene detect only)
155			double prev_mafd; ///< previous MAFD (scene detect only)
156			AVFrame *prev_picref; ///< previous frame (scene detect only)
157			double select;
158			int select_out; ///< mark the selected output pad index
159			int nb_outputs;
160			} SelectContext;
161
162			#define OFFSET(x) offsetof(SelectContext, x)
163			#define DEFINE_OPTIONS(filt_name, FLAGS) \
164			static const AVOption filt_name##_options[] = { \
165			{ "expr", "set an expression to use for selecting frames", OFFSET(expr_str), AV_OPT_TYPE_STRING, { .str = "1" }, .flags=FLAGS }, \
166			{ "e", "set an expression to use for selecting frames", OFFSET(expr_str), AV_OPT_TYPE_STRING, { .str = "1" }, .flags=FLAGS }, \
167			{ "outputs", "set the number of outputs", OFFSET(nb_outputs), AV_OPT_TYPE_INT, {.i64 = 1}, 1, INT_MAX, .flags=FLAGS }, \
168			{ "n", "set the number of outputs", OFFSET(nb_outputs), AV_OPT_TYPE_INT, {.i64 = 1}, 1, INT_MAX, .flags=FLAGS }, \
169			{ NULL } \
170			}
171
172			static int request_frame(AVFilterLink *outlink);
173
174		3	static av_cold int init(AVFilterContext *ctx)
175			{
176		3	SelectContext *select = ctx->priv;
177			int i, ret;
178
179	✗✓	3	if ((ret = av_expr_parse(&select->expr, select->expr_str,
180			var_names, NULL, NULL, NULL, NULL, 0, ctx)) < 0) {
181			av_log(ctx, AV_LOG_ERROR, "Error while parsing expression '%s'\n",
182			select->expr_str);
183			return ret;
184			}
185		3	select->do_scene_detect = !!strstr(select->expr_str, "scene");
186
187	✓✓	6	for (i = 0; i < select->nb_outputs; i++) {
188		3	AVFilterPad pad = { 0 };
189
190		3	pad.name = av_asprintf("output%d", i);
191	✗✓	3	if (!pad.name)
192			return AVERROR(ENOMEM);
193		3	pad.type = ctx->filter->inputs[0].type;
194		3	pad.request_frame = request_frame;
195	✗✓	3	if ((ret = ff_insert_outpad(ctx, i, &pad)) < 0) {
196			av_freep(&pad.name);
197			return ret;
198			}
199			}
200
201		3	return 0;
202			}
203
204			#define INTERLACE_TYPE_P 0
205			#define INTERLACE_TYPE_T 1
206			#define INTERLACE_TYPE_B 2
207
208		3	static int config_input(AVFilterLink *inlink)
209			{
210		3	SelectContext *select = inlink->dst->priv;
211		3	const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
212		9	int is_yuv = !(desc->flags & AV_PIX_FMT_FLAG_RGB) &&
213	✓✗✓✓	5	(desc->flags & AV_PIX_FMT_FLAG_PLANAR) &&
214	✓✗	2	desc->nb_components >= 3;
215
216		3	select->bitdepth = desc->comp[0].depth;
217	✓✓	3	select->nb_planes = is_yuv ? 1 : av_pix_fmt_count_planes(inlink->format);
218
219	✓✓	6	for (int plane = 0; plane < select->nb_planes; plane++) {
220		3	ptrdiff_t line_size = av_image_get_linesize(inlink->format, inlink->w, plane);
221		3	int vsub = desc->log2_chroma_h;
222
223		3	select->width[plane] = line_size >> (select->bitdepth > 8);
224	✓✗✗✓	3	select->height[plane] = plane == 1 \|\| plane == 2 ? AV_CEIL_RSHIFT(inlink->h, vsub) : inlink->h;
225			}
226
227		3	select->var_values[VAR_N] = 0.0;
228		3	select->var_values[VAR_SELECTED_N] = 0.0;
229
230		3	select->var_values[VAR_TB] = av_q2d(inlink->time_base);
231
232		3	select->var_values[VAR_PREV_PTS] = NAN;
233		3	select->var_values[VAR_PREV_SELECTED_PTS] = NAN;
234		3	select->var_values[VAR_PREV_SELECTED_T] = NAN;
235		3	select->var_values[VAR_PREV_T] = NAN;
236		3	select->var_values[VAR_START_PTS] = NAN;
237		3	select->var_values[VAR_START_T] = NAN;
238
239		3	select->var_values[VAR_I] = AV_PICTURE_TYPE_I;
240		3	select->var_values[VAR_P] = AV_PICTURE_TYPE_P;
241		3	select->var_values[VAR_B] = AV_PICTURE_TYPE_B;
242		3	select->var_values[VAR_SI] = AV_PICTURE_TYPE_SI;
243		3	select->var_values[VAR_SP] = AV_PICTURE_TYPE_SP;
244		3	select->var_values[VAR_BI] = AV_PICTURE_TYPE_BI;
245		3	select->var_values[VAR_PICT_TYPE_I] = AV_PICTURE_TYPE_I;
246		3	select->var_values[VAR_PICT_TYPE_P] = AV_PICTURE_TYPE_P;
247		3	select->var_values[VAR_PICT_TYPE_B] = AV_PICTURE_TYPE_B;
248		3	select->var_values[VAR_PICT_TYPE_SI] = AV_PICTURE_TYPE_SI;
249		3	select->var_values[VAR_PICT_TYPE_SP] = AV_PICTURE_TYPE_SP;
250		3	select->var_values[VAR_PICT_TYPE_BI] = AV_PICTURE_TYPE_BI;
251
252		3	select->var_values[VAR_INTERLACE_TYPE_P] = INTERLACE_TYPE_P;
253		3	select->var_values[VAR_INTERLACE_TYPE_T] = INTERLACE_TYPE_T;
254		3	select->var_values[VAR_INTERLACE_TYPE_B] = INTERLACE_TYPE_B;
255
256		3	select->var_values[VAR_PICT_TYPE] = NAN;
257		3	select->var_values[VAR_INTERLACE_TYPE] = NAN;
258		3	select->var_values[VAR_SCENE] = NAN;
259		3	select->var_values[VAR_CONSUMED_SAMPLES_N] = NAN;
260		3	select->var_values[VAR_SAMPLES_N] = NAN;
261
262		3	select->var_values[VAR_SAMPLE_RATE] =
263	✗✓	3	inlink->type == AVMEDIA_TYPE_AUDIO ? inlink->sample_rate : NAN;
264
265	✓✓	3	if (CONFIG_SELECT_FILTER && select->do_scene_detect) {
266	✓✗	1	select->sad = ff_scene_sad_get_fn(select->bitdepth == 8 ? 8 : 16);
267	✗✓	1	if (!select->sad)
268			return AVERROR(EINVAL);
269			}
270		3	return 0;
271			}
272
273		1308	static double get_scene_score(AVFilterContext ctx, AVFrame frame)
274			{
275		1308	double ret = 0;
276		1308	SelectContext *select = ctx->priv;
277		1308	AVFrame *prev_picref = select->prev_picref;
278
279	✓✓	1308	if (prev_picref &&
280	✓✗	1307	frame->height == prev_picref->height &&
281	✓✗	1307	frame->width == prev_picref->width) {
282		1307	uint64_t sad = 0;
283			double mafd, diff;
284		1307	uint64_t count = 0;
285
286	✓✓	2614	for (int plane = 0; plane < select->nb_planes; plane++) {
287			uint64_t plane_sad;
288		1307	select->sad(prev_picref->data[plane], prev_picref->linesize[plane],
289		1307	frame->data[plane], frame->linesize[plane],
290			select->width[plane], select->height[plane], &plane_sad);
291		1307	sad += plane_sad;
292		1307	count += select->width[plane] * select->height[plane];
293			}
294
295		1307	emms_c();
296		1307	mafd = (double)sad / count / (1ULL << (select->bitdepth - 8));
297		1307	diff = fabs(mafd - select->prev_mafd);
298	✓✓	1307	ret = av_clipf(FFMIN(mafd, diff) / 100., 0, 1);
299		1307	select->prev_mafd = mafd;
300		1307	av_frame_free(&prev_picref);
301			}
302		1308	select->prev_picref = av_frame_clone(frame);
303		1308	return ret;
304			}
305
306		1408	static double get_concatdec_select(AVFrame *frame, int64_t pts)
307			{
308		1408	AVDictionary *metadata = frame->metadata;
309		1408	AVDictionaryEntry *start_time_entry = av_dict_get(metadata, "lavf.concatdec.start_time", NULL, 0);
310		1408	AVDictionaryEntry *duration_entry = av_dict_get(metadata, "lavf.concatdec.duration", NULL, 0);
311	✗✓	1408	if (start_time_entry) {
312			int64_t start_time = strtoll(start_time_entry->value, NULL, 10);
313			if (pts >= start_time) {
314			if (duration_entry) {
315			int64_t duration = strtoll(duration_entry->value, NULL, 10);
316			if (pts < start_time + duration)
317			return -1;
318			else
319			return 0;
320			}
321			return -1;
322			}
323			return 0;
324			}
325		1408	return NAN;
326			}
327
328		1408	static void select_frame(AVFilterContext ctx, AVFrame frame)
329			{
330		1408	SelectContext *select = ctx->priv;
331		1408	AVFilterLink *inlink = ctx->inputs[0];
332			double res;
333
334	✓✓	1408	if (isnan(select->var_values[VAR_START_PTS]))
335	✓✗	3	select->var_values[VAR_START_PTS] = TS2D(frame->pts);
336	✓✓	1408	if (isnan(select->var_values[VAR_START_T]))
337	✓✗	3	select->var_values[VAR_START_T] = TS2D(frame->pts) * av_q2d(inlink->time_base);
338
339		1408	select->var_values[VAR_N ] = inlink->frame_count_out;
340	✓✗	1408	select->var_values[VAR_PTS] = TS2D(frame->pts);
341	✓✗	1408	select->var_values[VAR_T ] = TS2D(frame->pts) * av_q2d(inlink->time_base);
342	✓✓	1408	select->var_values[VAR_POS] = frame->pkt_pos == -1 ? NAN : frame->pkt_pos;
343		1408	select->var_values[VAR_KEY] = frame->key_frame;
344		1408	select->var_values[VAR_CONCATDEC_SELECT] = get_concatdec_select(frame, av_rescale_q(frame->pts, inlink->time_base, AV_TIME_BASE_Q));
345
346	✗✓✗	1408	switch (inlink->type) {
347			case AVMEDIA_TYPE_AUDIO:
348			select->var_values[VAR_SAMPLES_N] = frame->nb_samples;
349			break;
350
351		1408	case AVMEDIA_TYPE_VIDEO:
352		1408	select->var_values[VAR_INTERLACE_TYPE] =
353	✗✓	1408	!frame->interlaced_frame ? INTERLACE_TYPE_P :
354			frame->top_field_first ? INTERLACE_TYPE_T : INTERLACE_TYPE_B;
355		1408	select->var_values[VAR_PICT_TYPE] = frame->pict_type;
356	✓✓	1408	if (select->do_scene_detect) {
357			char buf[32];
358		1308	select->var_values[VAR_SCENE] = get_scene_score(ctx, frame);
359			// TODO: document metadata
360		1308	snprintf(buf, sizeof(buf), "%f", select->var_values[VAR_SCENE]);
361		1308	av_dict_set(&frame->metadata, "lavfi.scene_score", buf, 0);
362			}
363		1408	break;
364			}
365
366		1408	select->select = res = av_expr_eval(select->expr, select->var_values, NULL);
367		1408	av_log(inlink->dst, AV_LOG_DEBUG,
368			"n:%f pts:%f t:%f key:%d",
369			select->var_values[VAR_N],
370			select->var_values[VAR_PTS],
371			select->var_values[VAR_T],
372			frame->key_frame);
373
374	✓✗✗	1408	switch (inlink->type) {
375		1408	case AVMEDIA_TYPE_VIDEO:
376		1408	av_log(inlink->dst, AV_LOG_DEBUG, " interlace_type:%c pict_type:%c scene:%f",
377	✗✓	1408	(!frame->interlaced_frame) ? 'P' :
378			frame->top_field_first ? 'T' : 'B',
379		1408	av_get_picture_type_char(frame->pict_type),
380			select->var_values[VAR_SCENE]);
381		1408	break;
382			case AVMEDIA_TYPE_AUDIO:
383			av_log(inlink->dst, AV_LOG_DEBUG, " samples_n:%d consumed_samples_n:%f",
384			frame->nb_samples,
385			select->var_values[VAR_CONSUMED_SAMPLES_N]);
386			break;
387			}
388
389	✓✓	1408	if (res == 0) {
390		1355	select->select_out = -1; /* drop */
391	✓✗✗✓	53	} else if (isnan(res) \|\| res < 0) {
392			select->select_out = 0; /* first output */
393			} else {
394	✗✓	53	select->select_out = FFMIN(ceilf(res)-1, select->nb_outputs-1); /* other outputs */
395			}
396
397		1408	av_log(inlink->dst, AV_LOG_DEBUG, " -> select:%f select_out:%d\n", res, select->select_out);
398
399	✓✓	1408	if (res) {
400		53	select->var_values[VAR_PREV_SELECTED_N] = select->var_values[VAR_N];
401		53	select->var_values[VAR_PREV_SELECTED_PTS] = select->var_values[VAR_PTS];
402		53	select->var_values[VAR_PREV_SELECTED_T] = select->var_values[VAR_T];
403		53	select->var_values[VAR_SELECTED_N] += 1.0;
404	✗✓	53	if (inlink->type == AVMEDIA_TYPE_AUDIO)
405			select->var_values[VAR_CONSUMED_SAMPLES_N] += frame->nb_samples;
406			}
407
408		1408	select->var_values[VAR_PREV_PTS] = select->var_values[VAR_PTS];
409		1408	select->var_values[VAR_PREV_T] = select->var_values[VAR_T];
410		1408	}
411
412		1408	static int filter_frame(AVFilterLink inlink, AVFrame frame)
413			{
414		1408	AVFilterContext *ctx = inlink->dst;
415		1408	SelectContext *select = ctx->priv;
416
417		1408	select_frame(ctx, frame);
418	✓✓	1408	if (select->select)
419		53	return ff_filter_frame(ctx->outputs[select->select_out], frame);
420
421		1355	av_frame_free(&frame);
422		1355	return 0;
423			}
424
425		1412	static int request_frame(AVFilterLink *outlink)
426			{
427		1412	AVFilterLink *inlink = outlink->src->inputs[0];
428		1412	int ret = ff_request_frame(inlink);
429		1412	return ret;
430			}
431
432		3	static av_cold void uninit(AVFilterContext *ctx)
433			{
434		3	SelectContext *select = ctx->priv;
435			int i;
436
437		3	av_expr_free(select->expr);
438		3	select->expr = NULL;
439
440	✓✓	6	for (i = 0; i < ctx->nb_outputs; i++)
441		3	av_freep(&ctx->output_pads[i].name);
442
443	✓✓	3	if (select->do_scene_detect) {
444		1	av_frame_free(&select->prev_picref);
445			}
446		3	}
447
448			#if CONFIG_ASELECT_FILTER
449
450			DEFINE_OPTIONS(aselect, AV_OPT_FLAG_AUDIO_PARAM\|AV_OPT_FLAG_FILTERING_PARAM);
451			AVFILTER_DEFINE_CLASS(aselect);
452
453			static av_cold int aselect_init(AVFilterContext *ctx)
454			{
455			SelectContext *select = ctx->priv;
456			int ret;
457
458			if ((ret = init(ctx)) < 0)
459			return ret;
460
461			if (select->do_scene_detect) {
462			av_log(ctx, AV_LOG_ERROR, "Scene detection is ignored in aselect filter\n");
463			return AVERROR(EINVAL);
464			}
465
466			return 0;
467			}
468
469			static const AVFilterPad avfilter_af_aselect_inputs[] = {
470			{
471			.name = "default",
472			.type = AVMEDIA_TYPE_AUDIO,
473			.config_props = config_input,
474			.filter_frame = filter_frame,
475			},
476			{ NULL }
477			};
478
479			AVFilter ff_af_aselect = {
480			.name = "aselect",
481			.description = NULL_IF_CONFIG_SMALL("Select audio frames to pass in output."),
482			.init = aselect_init,
483			.uninit = uninit,
484			.priv_size = sizeof(SelectContext),
485			.inputs = avfilter_af_aselect_inputs,
486			.priv_class = &aselect_class,
487			.flags = AVFILTER_FLAG_DYNAMIC_OUTPUTS,
488			};
489			#endif /* CONFIG_ASELECT_FILTER */
490
491			#if CONFIG_SELECT_FILTER
492
493		3	static int query_formats(AVFilterContext *ctx)
494			{
495		3	SelectContext *select = ctx->priv;
496
497	✓✓	3	if (!select->do_scene_detect) {
498		2	return ff_default_query_formats(ctx);
499			} else {
500			int ret;
501			static const enum AVPixelFormat pix_fmts[] = {
502			AV_PIX_FMT_RGB24, AV_PIX_FMT_BGR24, AV_PIX_FMT_RGBA,
503			AV_PIX_FMT_ABGR, AV_PIX_FMT_BGRA, AV_PIX_FMT_GRAY8,
504			AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUVJ420P,
505			AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUVJ422P,
506			AV_PIX_FMT_YUV420P10,
507			AV_PIX_FMT_NONE
508			};
509		1	AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);
510
511	✗✓	1	if (!fmts_list)
512			return AVERROR(ENOMEM);
513		1	ret = ff_set_common_formats(ctx, fmts_list);
514	✗✓	1	if (ret < 0)
515			return ret;
516			}
517		1	return 0;
518			}
519
520			DEFINE_OPTIONS(select, AV_OPT_FLAG_VIDEO_PARAM\|AV_OPT_FLAG_FILTERING_PARAM);
521			AVFILTER_DEFINE_CLASS(select);
522
523		3	static av_cold int select_init(AVFilterContext *ctx)
524			{
525			int ret;
526
527	✗✓	3	if ((ret = init(ctx)) < 0)
528			return ret;
529
530		3	return 0;
531			}
532
533			static const AVFilterPad avfilter_vf_select_inputs[] = {
534			{
535			.name = "default",
536			.type = AVMEDIA_TYPE_VIDEO,
537			.config_props = config_input,
538			.filter_frame = filter_frame,
539			},
540			{ NULL }
541			};
542
543			AVFilter ff_vf_select = {
544			.name = "select",
545			.description = NULL_IF_CONFIG_SMALL("Select video frames to pass in output."),
546			.init = select_init,
547			.uninit = uninit,
548			.query_formats = query_formats,
549			.priv_size = sizeof(SelectContext),
550			.priv_class = &select_class,
551			.inputs = avfilter_vf_select_inputs,
552			.flags = AVFILTER_FLAG_DYNAMIC_OUTPUTS,
553			};
554			#endif /* CONFIG_SELECT_FILTER */