Marton Balint:
> Previously only 1:1 bitstream filters were supported, the end of the stream
> was
> not signalled to the bitstream filters and time base changes were ignored.
>
> Signed-off-by: Marton Balint <c...@passwd.hu>
> ---
> libavformat/internal.h | 1 +
> libavformat/mux.c | 128
> ++++++++++++++++++++++++++++++++++---------------
> 2 files changed, 91 insertions(+), 38 deletions(-)
>
> diff --git a/libavformat/internal.h b/libavformat/internal.h
> index 332477a532..45aeef717a 100644
> --- a/libavformat/internal.h
> +++ b/libavformat/internal.h
> @@ -158,6 +158,7 @@ struct AVStreamInternal {
> */
> AVBSFContext **bsfcs;
> int nb_bsfcs;
> + int bsfcs_idx;
>
> /**
> * Whether or not check_bitstream should still be run on each packet
> diff --git a/libavformat/mux.c b/libavformat/mux.c
> index 8c2d6a8060..3054ab8644 100644
> --- a/libavformat/mux.c
> +++ b/libavformat/mux.c
> @@ -840,14 +840,48 @@ static int prepare_input_packet(AVFormatContext *s,
> AVPacket *pkt)
> return 0;
> }
>
> -static int do_packet_auto_bsf(AVFormatContext *s, AVPacket *pkt) {
> - AVStream *st = s->streams[pkt->stream_index];
> - int i, ret;
> +static int auto_bsf_receive_packet(AVFormatContext *s, AVStream *st,
> AVPacket *pkt)
> +{
> + AVStreamInternal *sti = st->internal;
> + int ret = AVERROR(EAGAIN);
> + int eof = 0;
> +
> + while (sti->bsfcs_idx) {
> + /* get a packet from the previous filter up the chain */
> + ret = av_bsf_receive_packet(sti->bsfcs[sti->bsfcs_idx - 1], pkt);
> + if (ret == AVERROR(EAGAIN)) {
> + sti->bsfcs_idx--;
> + continue;
> + } else if (ret == AVERROR_EOF) {
> + eof = 1;
> + } else if (ret < 0)
> + break;
> +
> + /* send it to the next filter down the chain */
> + if (sti->bsfcs_idx < sti->nb_bsfcs) {
> + ret = av_bsf_send_packet(sti->bsfcs[sti->bsfcs_idx], eof ? NULL
> : pkt);
> + av_assert2(ret != AVERROR(EAGAIN));
> + if (ret < 0)
> + break;
> + sti->bsfcs_idx++;
> + eof = 0;
> + } else if (eof) {
> + break;
> + } else {
> + return 0;
> + }
> + }
Would it actually be possible to simplify this by using the
av_bsf_list-API (right now code like this exists here and in ffmpeg.c
and probably at even more places)? The problem I see with this is that
one cannot send a packet for filtering to an unfinished bsf list whereas
the current code has this capability (although AFAIK no one uses it; if
I am not mistaken, all bsf chains that are automatically inserted
consist of exactly one bsf).
(This actually brings up a question: If check_bitstream returns that
more packets from this stream need to be checked, the current packet is
currently sent to the bsf-list as-is; it would not be sent to any bsf
that get added to the list later. Makes me wonder whether this is
actually a problem.)
- Andreas
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
