fftools/ffmpeg: add an option for writing pre-muxing stats

Analogous to -enc_stats*, but happens right before muxing. Useful
because bitstream filters and the sync queue can modify packets after
encoding and before muxing. Also has access to the muxing timebase.

Changelog

@@ -32,7 +32,8 @@ version <next>:
 - WADY DPCM decoder and demuxer
 - CBD2 DPCM decoder
 - ssim360 video filter
-- ffmpeg CLI new options: -enc_stats_pre[_fmt], -enc_stats_post[_fmt]
+- ffmpeg CLI new options: -enc_stats_pre[_fmt], -enc_stats_post[_fmt],
+  -stats_mux_pre[_fmt]
 - hstack_vaapi, vstack_vaapi and xstack_vaapi filters
 - XMD ADPCM decoder and demuxer
 - media100 to mjpegb bsf

doc/ffmpeg.texi

@@ -2063,14 +2063,17 @@ or invalid output files.
 
 @item -enc_stats_pre[:@var{stream_specifier}] @var{path} (@emph{output,per-stream})
 @item -enc_stats_post[:@var{stream_specifier}] @var{path} (@emph{output,per-stream})
+@item -stats_mux_pre[:@var{stream_specifier}] @var{path} (@emph{output,per-stream})
 Write per-frame encoding information about the matching streams into the file
 given by @var{path}.
 
 @option{-enc_stats_pre} writes information about raw video or audio frames right
 before they are sent for encoding, while @option{-enc_stats_post} writes
-information about encoded packets as they are received from the encoder. Every
+information about encoded packets as they are received from the encoder.
-frame or packet produces one line in the specified file. The format of this line
+@option{-stats_mux_pre} writes information about packets just as they are about to
-is controlled by @option{-enc_stats_pre_fmt} / @option{-enc_stats_post_fmt}.
+be sent to the muxer. Every frame or packet produces one line in the specified
+file. The format of this line is controlled by @option{-enc_stats_pre_fmt} /
+@option{-enc_stats_post_fmt} / @option{-stats_mux_pre_fmt}.
 
 When stats for multiple streams are written into a single file, the lines
 corresponding to different streams will be interleaved. The precise order of
@@ -2079,8 +2082,9 @@ different invocations of the program, even with the same options.
 
 @item -enc_stats_pre_fmt[:@var{stream_specifier}] @var{format_spec} (@emph{output,per-stream})
 @item -enc_stats_post_fmt[:@var{stream_specifier}] @var{format_spec} (@emph{output,per-stream})
+@item -stats_mux_pre_fmt[:@var{stream_specifier}] @var{format_spec} (@emph{output,per-stream})
 Specify the format for the lines written with @option{-enc_stats_pre} /
-@option{-enc_stats_post}.
+@option{-enc_stats_post} / @option{-stats_mux_pre}.
 
 @var{format_spec} is a string that may contain directives of the form
 @var{@{fmt@}}. @var{format_spec} is backslash-escaped --- use \@{, \@}, and \\
@@ -2097,6 +2101,7 @@ Index of the output stream in the file.
 @item n
 Frame number. Pre-encoding: number of frames sent to the encoder so far.
 Post-encoding: number of packets received from the encoder so far.
+Muxing: number of packets submitted to the muxer for this stream so far.
 
 @item ni
 Input frame number. Index of the input frame (i.e. output by a decoder) that

fftools/ffmpeg.c

@@ -808,8 +808,9 @@ static void update_video_stats(OutputStream *ost, const AVPacket *pkt, int write
     fprintf(vstats_file, "type= %c\n", av_get_picture_type_char(ost->pict_type));
 }
 
-static void enc_stats_write(OutputStream *ost, EncStats *es,
+void enc_stats_write(OutputStream *ost, EncStats *es,
-                            const AVFrame *frame, const AVPacket *pkt)
+                     const AVFrame *frame, const AVPacket *pkt,
+                     uint64_t frame_num)
 {
     AVIOContext *io = es->io;
     AVRational   tb = frame ? frame->time_base : pkt->time_base;
@@ -840,12 +841,12 @@ static void enc_stats_write(OutputStream *ost, EncStats *es,
         case ENC_STATS_PTS_TIME:        avio_printf(io, "%g",       pts * av_q2d(tb));              continue;
         case ENC_STATS_PTS_TIME_IN:     avio_printf(io, "%g",       ptsi == INT64_MAX ?
                                                                     INFINITY : ptsi * av_q2d(tbi)); continue;
+        case ENC_STATS_FRAME_NUM:       avio_printf(io, "%"PRIu64,  frame_num);                     continue;
         case ENC_STATS_FRAME_NUM_IN:    avio_printf(io, "%"PRIu64,  fd ? fd->idx : -1);             continue;
         }
 
         if (frame) {
             switch (c->type) {
-            case ENC_STATS_FRAME_NUM:   avio_printf(io, "%"PRIu64,  ost->frames_encoded);           continue;
             case ENC_STATS_SAMPLE_NUM:  avio_printf(io, "%"PRIu64,  ost->samples_encoded);          continue;
             case ENC_STATS_NB_SAMPLES:  avio_printf(io, "%d",       frame->nb_samples);             continue;
             default: av_assert0(0);
@@ -855,7 +856,6 @@ static void enc_stats_write(OutputStream *ost, EncStats *es,
             case ENC_STATS_DTS:         avio_printf(io, "%"PRId64,  pkt->dts);                      continue;
             case ENC_STATS_DTS_TIME:    avio_printf(io, "%g",       pkt->dts * av_q2d(tb));         continue;
             case ENC_STATS_PKT_SIZE:    avio_printf(io, "%d",       pkt->size);                     continue;
-            case ENC_STATS_FRAME_NUM:   avio_printf(io, "%"PRIu64,  ost->packets_encoded);          continue;
             case ENC_STATS_BITRATE: {
                 double duration = FFMAX(pkt->duration, 1) * av_q2d(tb);
                 avio_printf(io, "%g",  8.0 * pkt->size / duration);
@@ -884,7 +884,8 @@ static int encode_frame(OutputFile *of, OutputStream *ost, AVFrame *frame)
 
     if (frame) {
         if (ost->enc_stats_pre.io)
-            enc_stats_write(ost, &ost->enc_stats_pre, frame, NULL);
+            enc_stats_write(ost, &ost->enc_stats_pre, frame, NULL,
+                            ost->frames_encoded);
 
         ost->frames_encoded++;
         ost->samples_encoded += frame->nb_samples;
@@ -932,7 +933,8 @@ static int encode_frame(OutputFile *of, OutputStream *ost, AVFrame *frame)
         if (enc->codec_type == AVMEDIA_TYPE_VIDEO)
             update_video_stats(ost, pkt, !!vstats_filename);
         if (ost->enc_stats_post.io)
-            enc_stats_write(ost, &ost->enc_stats_post, NULL, pkt);
+            enc_stats_write(ost, &ost->enc_stats_post, NULL, pkt,
+                            ost->packets_encoded);
 
         if (debug_ts) {
             av_log(ost, AV_LOG_INFO, "encoder -> type:%s "

fftools/ffmpeg.h

@@ -258,10 +258,14 @@ typedef struct OptionsContext {
     int        nb_enc_stats_pre;
     SpecifierOpt *enc_stats_post;
     int        nb_enc_stats_post;
+    SpecifierOpt *mux_stats;
+    int        nb_mux_stats;
     SpecifierOpt *enc_stats_pre_fmt;
     int        nb_enc_stats_pre_fmt;
     SpecifierOpt *enc_stats_post_fmt;
     int        nb_enc_stats_post_fmt;
+    SpecifierOpt *mux_stats_fmt;
+    int        nb_mux_stats_fmt;
 } OptionsContext;
 
 typedef struct InputFilter {
@@ -789,6 +793,10 @@ int ifilter_parameters_from_frame(InputFilter *ifilter, const AVFrame *frame);
 
 int ffmpeg_parse_options(int argc, char **argv);
 
+void enc_stats_write(OutputStream *ost, EncStats *es,
+                     const AVFrame *frame, const AVPacket *pkt,
+                     uint64_t frame_num);
+
 HWDevice *hw_device_get_by_name(const char *name);
 int hw_device_init_from_string(const char *arg, HWDevice **dev);
 void hw_device_free_all(void);

fftools/ffmpeg_mux.c

@@ -64,6 +64,7 @@ static int write_packet(Muxer *mux, OutputStream *ost, AVPacket *pkt)
     AVFormatContext *s = mux->fc;
     AVStream *st = ost->st;
     int64_t fs;
+    uint64_t frame_num;
     int ret;
 
     fs = filesize(s->pb);
@@ -128,7 +129,7 @@ static int write_packet(Muxer *mux, OutputStream *ost, AVPacket *pkt)
     ms->last_mux_dts = pkt->dts;
 
     ost->data_size_mux += pkt->size;
-    atomic_fetch_add(&ost->packets_written, 1);
+    frame_num = atomic_fetch_add(&ost->packets_written, 1);
 
     pkt->stream_index = ost->index;
 
@@ -143,6 +144,9 @@ static int write_packet(Muxer *mux, OutputStream *ost, AVPacket *pkt)
               );
     }
 
+    if (ms->stats.io)
+        enc_stats_write(ost, &ms->stats, NULL, pkt, frame_num);
+
     ret = av_interleaved_write_frame(s, pkt);
     if (ret < 0) {
         print_error("av_interleaved_write_frame()", ret);
@@ -688,6 +692,10 @@ static void ost_free(OutputStream **post)
         av_freep(&ost->enc_stats_post.components[i].str);
     av_freep(&ost->enc_stats_post.components);
 
+    for (int i = 0; i < ms->stats.nb_components; i++)
+        av_freep(&ms->stats.components[i].str);
+    av_freep(&ms->stats.components);
+
     av_freep(post);
 }
 

fftools/ffmpeg_mux.h

@@ -45,6 +45,8 @@ typedef struct MuxStream {
 
     AVBSFContext *bsf_ctx;
 
+    EncStats stats;
+
     int64_t max_frames;
 
     /*

fftools/ffmpeg_mux_init.c

@@ -57,8 +57,10 @@ static const char *const opt_name_disposition[]               = {"disposition",
 static const char *const opt_name_enc_time_bases[]            = {"enc_time_base", NULL};
 static const char *const opt_name_enc_stats_pre[]             = {"enc_stats_pre", NULL};
 static const char *const opt_name_enc_stats_post[]            = {"enc_stats_post", NULL};
+static const char *const opt_name_mux_stats[]                 = {"mux_stats", NULL};
 static const char *const opt_name_enc_stats_pre_fmt[]         = {"enc_stats_pre_fmt", NULL};
 static const char *const opt_name_enc_stats_post_fmt[]        = {"enc_stats_post_fmt", NULL};
+static const char *const opt_name_mux_stats_fmt[]             = {"mux_stats_fmt", NULL};
 static const char *const opt_name_filters[]                   = {"filter", "af", "vf", NULL};
 static const char *const opt_name_filter_scripts[]            = {"filter_script", NULL};
 static const char *const opt_name_fix_sub_duration_heartbeat[] = {"fix_sub_duration_heartbeat", NULL};
@@ -262,7 +264,7 @@ static int unescape(char **pdst, size_t *dst_len,
     return 0;
 }
 
-static int enc_stats_init(OutputStream *ost, int pre,
+static int enc_stats_init(OutputStream *ost, EncStats *es, int pre,
                           const char *path, const char *fmt_spec)
 {
     static const struct {
@@ -290,7 +292,6 @@ static int enc_stats_init(OutputStream *ost, int pre,
         { ENC_STATS_BITRATE,        "br",       0, 1            },
         { ENC_STATS_AVG_BITRATE,    "abr",      0, 1            },
     };
-    EncStats *es = pre ? &ost->enc_stats_pre : &ost->enc_stats_post;
     const char *next = fmt_spec;
 
     int ret;
@@ -479,7 +480,7 @@ static OutputStream *new_output_stream(Muxer *mux, const OptionsContext *o,
         AVCodecContext *enc = ost->enc_ctx;
         AVIOContext *s = NULL;
         char *buf = NULL, *arg = NULL, *preset = NULL;
-        const char *enc_stats_pre = NULL, *enc_stats_post = NULL;
+        const char *enc_stats_pre = NULL, *enc_stats_post = NULL, *mux_stats = NULL;
 
         ost->encoder_opts = filter_codec_opts(o->g->codec_opts, enc->codec_id,
                                               oc, st, enc->codec);
@@ -518,7 +519,7 @@ static OutputStream *new_output_stream(Muxer *mux, const OptionsContext *o,
 
             MATCH_PER_STREAM_OPT(enc_stats_pre_fmt, str, format, oc, st);
 
-            ret = enc_stats_init(ost, 1, enc_stats_pre, format);
+            ret = enc_stats_init(ost, &ost->enc_stats_pre, 1, enc_stats_pre, format);
             if (ret < 0)
                 exit_program(1);
         }
@@ -530,7 +531,19 @@ static OutputStream *new_output_stream(Muxer *mux, const OptionsContext *o,
 
             MATCH_PER_STREAM_OPT(enc_stats_post_fmt, str, format, oc, st);
 
-            ret = enc_stats_init(ost, 0, enc_stats_post, format);
+            ret = enc_stats_init(ost, &ost->enc_stats_post, 0, enc_stats_post, format);
+            if (ret < 0)
+                exit_program(1);
+        }
+
+        MATCH_PER_STREAM_OPT(mux_stats, str, mux_stats, oc, st);
+        if (mux_stats &&
+            (type == AVMEDIA_TYPE_VIDEO || type == AVMEDIA_TYPE_AUDIO)) {
+            const char *format = "{fidx} {sidx} {n} {t}";
+
+            MATCH_PER_STREAM_OPT(mux_stats_fmt, str, format, oc, st);
+
+            ret = enc_stats_init(ost, &ms->stats, 0, mux_stats, format);
             if (ret < 0)
                 exit_program(1);
         }

fftools/ffmpeg_opt.c

@@ -1548,10 +1548,14 @@ const OptionDef options[] = {
         "write encoding stats before encoding" },
     { "enc_stats_post",     HAS_ARG | OPT_SPEC | OPT_EXPERT | OPT_OUTPUT | OPT_STRING, { .off = OFFSET(enc_stats_post)     },
         "write encoding stats after encoding" },
+    { "stats_mux_pre",      HAS_ARG | OPT_SPEC | OPT_EXPERT | OPT_OUTPUT | OPT_STRING, { .off = OFFSET(mux_stats)          },
+        "write packets stats before muxing" },
     { "enc_stats_pre_fmt",  HAS_ARG | OPT_SPEC | OPT_EXPERT | OPT_OUTPUT | OPT_STRING, { .off = OFFSET(enc_stats_pre_fmt)  },
         "format of the stats written with -enc_stats_pre" },
     { "enc_stats_post_fmt", HAS_ARG | OPT_SPEC | OPT_EXPERT | OPT_OUTPUT | OPT_STRING, { .off = OFFSET(enc_stats_post_fmt) },
         "format of the stats written with -enc_stats_post" },
+    { "stats_mux_pre_fmt",  HAS_ARG | OPT_SPEC | OPT_EXPERT | OPT_OUTPUT | OPT_STRING, { .off = OFFSET(mux_stats_fmt)      },
+        "format of the stats written with -stats_mux_pre" },
 
     /* video options */
     { "vframes",      OPT_VIDEO | HAS_ARG  | OPT_PERFILE | OPT_OUTPUT,           { .func_arg = opt_video_frames },