From: Nicolas Gaullier <nicolas.gaullier@cji.paris> To: ffmpeg-devel@ffmpeg.org Cc: Nicolas Gaullier <nicolas.gaullier@cji.paris> Subject: [FFmpeg-devel] [PATCH] avformat/mpegts: fix first NAL start code splited in two different packets Date: Fri, 2 Feb 2024 16:23:38 +0100 Message-ID: <20240202152338.609500-2-nicolas.gaullier@cji.paris> (raw) In-Reply-To: <20240202152338.609500-1-nicolas.gaullier@cji.paris> When PES are not aligned and a tiny nal-encoded frame is contained in a single TS packet, the first NAL startcode may be split by the PES boundary, making the packet unworkable. This patch shift the PES boundaries to avoid this. Signed-off-by: Nicolas Gaullier <nicolas.gaullier@cji.paris> --- libavformat/mpegts.c | 41 +++++++++++++++++++++++++++++++++++++++-- 1 file changed, 39 insertions(+), 2 deletions(-) diff --git a/libavformat/mpegts.c b/libavformat/mpegts.c index bef00c88e7..4b4cf82fb9 100644 --- a/libavformat/mpegts.c +++ b/libavformat/mpegts.c @@ -1149,6 +1149,7 @@ static int mpegts_push_data(MpegTSFilter *filter, PESContext *pes = filter->u.pes_filter.opaque; MpegTSContext *ts = pes->ts; const uint8_t *p; + int pes_align_shift = 0; int ret, len; if (!ts->pkt) @@ -1156,6 +1157,37 @@ static int mpegts_push_data(MpegTSFilter *filter, if (is_start) { if (pes->state == MPEGTS_PAYLOAD && pes->data_index > 0) { + if ((pes->st->codecpar->codec_id == AV_CODEC_ID_H264 + || pes->st->codecpar->codec_id == AV_CODEC_ID_HEVC) + && pes->data_index < TS_PACKET_SIZE - 4 - PES_HEADER_SIZE + && pes->data_index >= 4 + && buf_size >= 4 ) { + /* check/avoid spliting the start code + first byte of the first NAL unit in two different packets. + * this could happen with a tiny unaligned PES that fits in a single ts packet. */ + uint8_t *last_p_end = pes->buffer->data + pes->data_index - 4; + p = buf + PES_HEADER_SIZE + buf[PES_HEADER_SIZE - 1]; + if (last_p_end[3] == 0x00 && AV_RB24(p) == 0x000001) + pes_align_shift = 4; + else if (AV_RB16(last_p_end + 2)== 0x0000 && AV_RB16(p) == 0x0001) + pes_align_shift = 3; + else if (AV_RB24(last_p_end + 1)== 0x000000 && *p == 0x01) + pes_align_shift = 2; + else if (AV_RB32(last_p_end) == 0x00000001) + pes_align_shift = 1; + if (pes_align_shift) + { + last_p_end += 4; + if (pes_align_shift > 3) + *last_p_end++ = 0x00; + if (pes_align_shift > 2) + *last_p_end++ = 0x00; + if (pes_align_shift > 1) + *last_p_end++ = 0x01; + *last_p_end = *(p + pes_align_shift - 1); + pes->data_index += pes_align_shift; + buf_size -= pes_align_shift; + } + } ret = new_pes_packet(pes, ts->pkt); if (ret < 0) return ret; @@ -1301,6 +1333,7 @@ skip: /* we got the full header. We parse it and get the payload */ pes->state = MPEGTS_PAYLOAD; pes->data_index = 0; + p += pes_align_shift; if (pes->stream_type == 0x12 && buf_size > 0) { int sl_header_bytes = read_sl_header(pes, &pes->sl, p, buf_size); @@ -1409,9 +1442,13 @@ skip: pes->data_index += buf_size; /* emit complete packets with known packet size * decreases demuxer delay for infrequent packets like subtitles from - * a couple of seconds to milliseconds for properly muxed files. */ + * a couple of seconds to milliseconds for properly muxed files. + * disabled for video/NALs because at this point it could split/break the first NAL start code. + */ if (!ts->stop_parse && pes->PES_packet_length && - pes->pes_header_size + pes->data_index == pes->PES_packet_length + PES_START_SIZE) { + pes->pes_header_size + pes->data_index == pes->PES_packet_length + PES_START_SIZE && + pes->st->codecpar->codec_id != AV_CODEC_ID_H264 && + pes->st->codecpar->codec_id != AV_CODEC_ID_HEVC) { ts->stop_parse = 1; ret = new_pes_packet(pes, ts->pkt); pes->state = MPEGTS_SKIP; -- 2.30.2 _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
next prev parent reply other threads:[~2024-02-02 15:24 UTC|newest] Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top 2024-02-02 15:23 [FFmpeg-devel] [PATCH 0/1] " Nicolas Gaullier 2024-02-02 15:23 ` Nicolas Gaullier [this message] 2024-02-06 10:58 ` Nicolas Gaullier 2024-02-06 11:28 ` Kieran Kunhya 2024-02-06 11:38 ` Nicolas Gaullier 2024-02-06 12:15 ` Kieran Kunhya 2024-02-06 21:36 ` Marton Balint 2024-02-20 16:58 ` Nicolas Gaullier
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20240202152338.609500-2-nicolas.gaullier@cji.paris \ --to=nicolas.gaullier@cji.paris \ --cc=ffmpeg-devel@ffmpeg.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: link
Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel This inbox may be cloned and mirrored by anyone: git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git # If you have public-inbox 1.1+ installed, you may # initialize and index your mirror using the following commands: public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \ ffmpegdev@gitmailbox.com public-inbox-index ffmpegdev Example config snippet for mirrors. AGPL code for this site: git clone https://public-inbox.org/public-inbox.git