From 7d8431004a236c7f3ea600753989fad11fd2d6df Mon Sep 17 00:00:00 2001 From: Luca Barbato Date: Mon, 17 Sep 2012 02:48:02 +0200 Subject: [PATCH 1/5] matroskadec: properly support BlockDuration --- libavformat/matroskadec.c | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-) diff --git a/libavformat/matroskadec.c b/libavformat/matroskadec.c index c75df51bb4..0a35a875c1 100644 --- a/libavformat/matroskadec.c +++ b/libavformat/matroskadec.c @@ -1992,7 +1992,7 @@ static int matroska_parse_frame(MatroskaDemuxContext *matroska, static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, int size, int64_t pos, uint64_t cluster_time, - uint64_t duration, int is_keyframe, + uint64_t block_duration, int is_keyframe, int64_t cluster_pos) { uint64_t timecode = AV_NOPTS_VALUE; @@ -2002,7 +2002,7 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, int16_t block_time; uint32_t *lace_size = NULL; int n, flags, laces = 0; - uint64_t num; + uint64_t num, duration; if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) { av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n"); @@ -2021,8 +2021,6 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, st = track->stream; if (st->discard >= AVDISCARD_ALL) return res; - if (duration == AV_NOPTS_VALUE) - duration = track->default_duration / matroska->time_scale; block_time = AV_RB16(data); data += 2; @@ -2039,7 +2037,6 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, is_keyframe = 0; /* overlapping subtitles are not key frame */ if (is_keyframe) av_add_index_entry(st, cluster_pos, timecode, 0,0,AVINDEX_KEYFRAME); - track->end_timecode = FFMAX(track->end_timecode, timecode+duration); } if (matroska->skip_to_keyframe && track->type != MATROSKA_TRACK_TYPE_SUBTITLE) { @@ -2054,6 +2051,21 @@ static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, if (res) goto end; + if (block_duration != AV_NOPTS_VALUE) { + duration = block_duration / laces; + if (block_duration != duration * laces) { + av_log(matroska->ctx, AV_LOG_WARNING, + "Incorrect block_duration, possibly corrupted container"); + } + } else { + duration = track->default_duration / matroska->time_scale; + block_duration = duration * laces; + } + + if (timecode != AV_NOPTS_VALUE) + track->end_timecode = + FFMAX(track->end_timecode, timecode + block_duration); + for (n = 0; n < laces; n++) { if ((st->codec->codec_id == AV_CODEC_ID_RA_288 || st->codec->codec_id == AV_CODEC_ID_COOK || From d17d0ec8c2b432f0b680737de0b2aa03f2d4dd8d Mon Sep 17 00:00:00 2001 From: Yusuke Nakamura Date: Tue, 24 Apr 2012 08:51:29 +0900 Subject: [PATCH 2/5] mov: support random access point grouping Frames described by this grouping are the starter of a closed or an open GOP. This is useful for open GOP of H.264 stream which is not described by sync sample atom. Signed-off-by: Luca Barbato --- libavformat/isom.h | 7 ++++++ libavformat/mov.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 60 insertions(+) diff --git a/libavformat/isom.h b/libavformat/isom.h index a766326082..b191699711 100644 --- a/libavformat/isom.h +++ b/libavformat/isom.h @@ -87,6 +87,11 @@ typedef struct { unsigned flags; } MOVTrackExt; +typedef struct { + unsigned int count; + unsigned int index; +} MOVSbgp; + typedef struct MOVStreamContext { AVIOContext *pb; int ffindex; ///< AVStream index @@ -128,6 +133,8 @@ typedef struct MOVStreamContext { int has_palette; int64_t data_size; int64_t track_end; ///< used for dts generation in fragmented movie files + unsigned int rap_group_count; + MOVSbgp *rap_group; } MOVStreamContext; typedef struct MOVContext { diff --git a/libavformat/mov.c b/libavformat/mov.c index 56a075e0b0..63049f559f 100644 --- a/libavformat/mov.c +++ b/libavformat/mov.c @@ -1736,6 +1736,46 @@ static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom) return 0; } +static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom) +{ + AVStream *st; + MOVStreamContext *sc; + unsigned int i, entries; + uint8_t version; + uint32_t grouping_type; + + if (c->fc->nb_streams < 1) + return 0; + st = c->fc->streams[c->fc->nb_streams-1]; + sc = st->priv_data; + + version = avio_r8(pb); /* version */ + avio_rb24(pb); /* flags */ + grouping_type = avio_rl32(pb); + if (grouping_type != MKTAG( 'r','a','p',' ')) + return 0; /* only support 'rap ' grouping */ + if (version == 1) + avio_rb32(pb); /* grouping_type_parameter */ + + entries = avio_rb32(pb); + if (!entries) + return 0; + if (entries >= UINT_MAX / sizeof(*sc->rap_group)) + return AVERROR_INVALIDDATA; + sc->rap_group = av_malloc(entries * sizeof(*sc->rap_group)); + if (!sc->rap_group) + return AVERROR(ENOMEM); + + for (i = 0; i < entries && !pb->eof_reached; i++) { + sc->rap_group[i].count = avio_rb32(pb); /* sample_count */ + sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */ + } + + sc->rap_group_count = i; + + return pb->eof_reached ? AVERROR_EOF : 0; +} + static void mov_build_index(MOVContext *mov, AVStream *st) { MOVStreamContext *sc = st->priv_data; @@ -1770,6 +1810,9 @@ static void mov_build_index(MOVContext *mov, AVStream *st) unsigned int stts_sample = 0; unsigned int sample_size; unsigned int distance = 0; + unsigned int rap_group_index = 0; + unsigned int rap_group_sample = 0; + int rap_group_present = sc->rap_group_count && sc->rap_group; int key_off = (sc->keyframes && sc->keyframes[0] > 0) || (sc->stps_data && sc->stps_data[0] > 0); current_dts -= sc->dts_shift; @@ -1805,6 +1848,14 @@ static void mov_build_index(MOVContext *mov, AVStream *st) if (stps_index + 1 < sc->stps_count) stps_index++; } + if (rap_group_present && rap_group_index < sc->rap_group_count) { + if (sc->rap_group[rap_group_index].index > 0) + keyframe = 1; + if (++rap_group_sample == sc->rap_group[rap_group_index].count) { + rap_group_sample = 0; + rap_group_index++; + } + } if (keyframe) distance = 0; sample_size = sc->sample_size > 0 ? sc->sample_size : sc->sample_sizes[current_sample]; @@ -2054,6 +2105,7 @@ static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom) av_freep(&sc->keyframes); av_freep(&sc->stts_data); av_freep(&sc->stps_data); + av_freep(&sc->rap_group); return 0; } @@ -2497,6 +2549,7 @@ static const MOVParseTableEntry mov_default_parse_table[] = { { MKTAG('c','m','o','v'), mov_read_cmov }, { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */ { MKTAG('d','v','c','1'), mov_read_dvc1 }, +{ MKTAG('s','b','g','p'), mov_read_sbgp }, { 0, NULL } }; From 9bf41210a9a7511f44ea1d7fd2646430889a7c2b Mon Sep 17 00:00:00 2001 From: John Van Sickle Date: Wed, 19 Sep 2012 13:51:29 -0400 Subject: [PATCH 3/5] libx264: add support for nal-hrd, required for Blu-ray streams. Signed-off-by: Anton Khirnov --- libavcodec/libx264.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/libavcodec/libx264.c b/libavcodec/libx264.c index 91cac531ed..62815ced92 100644 --- a/libavcodec/libx264.c +++ b/libavcodec/libx264.c @@ -69,6 +69,7 @@ typedef struct X264Context { int direct_pred; int slice_max_size; char *stats; + int nal_hrd } X264Context; static void X264_log(void *p, int level, const char *fmt, va_list args) @@ -373,6 +374,9 @@ static av_cold int X264_init(AVCodecContext *avctx) if (x4->fastfirstpass) x264_param_apply_fastfirstpass(&x4->params); + if (x4->nal_hrd >= 0) + x4->params.i_nal_hrd = x4->nal_hrd; + if (x4->profile) if (x264_param_apply_profile(&x4->params, x4->profile) < 0) { av_log(avctx, AV_LOG_ERROR, "Error setting profile %s.\n", x4->profile); @@ -518,6 +522,11 @@ static const AVOption options[] = { { "auto", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_AUTO }, 0, 0, VE, "direct-pred" }, { "slice-max-size","Limit the size of each slice in bytes", OFFSET(slice_max_size),AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE }, { "stats", "Filename for 2 pass stats", OFFSET(stats), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE }, + { "nal-hrd", "Signal HRD information (requires vbv-bufsize; " + "cbr not allowed in .mp4)", OFFSET(nal_hrd), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "nal-hrd" }, + { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_NONE}, INT_MIN, INT_MAX, VE, "nal-hrd" }, + { "vbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_VBR}, INT_MIN, INT_MAX, VE, "nal-hrd" }, + { "cbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_CBR}, INT_MIN, INT_MAX, VE, "nal-hrd" }, { NULL }, }; From 2d1a1a7f62623ff153184dc75dd9559d93b76619 Mon Sep 17 00:00:00 2001 From: Anton Khirnov Date: Sun, 2 Sep 2012 00:30:41 +0200 Subject: [PATCH 4/5] mp3dec: use named constants for Xing header flags --- libavformat/mp3dec.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/libavformat/mp3dec.c b/libavformat/mp3dec.c index d81d0fc97f..ffc1c35a60 100644 --- a/libavformat/mp3dec.c +++ b/libavformat/mp3dec.c @@ -29,6 +29,9 @@ #include "id3v1.h" #include "libavcodec/mpegaudiodecheader.h" +#define XING_FLAG_FRAMES 0x01 +#define XING_FLAG_SIZE 0x02 + /* mp3 read */ static int mp3_read_probe(AVProbeData *p) @@ -117,9 +120,9 @@ static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base) v = avio_rb32(s->pb); if(v == MKBETAG('X', 'i', 'n', 'g') || v == MKBETAG('I', 'n', 'f', 'o')) { v = avio_rb32(s->pb); - if(v & 0x1) + if(v & XING_FLAG_FRAMES) frames = avio_rb32(s->pb); - if(v & 0x2) + if(v & XING_FLAG_SIZE) size = avio_rb32(s->pb); } From 50d1f4437be88a4b7e412e90d71153cae68017cc Mon Sep 17 00:00:00 2001 From: Anton Khirnov Date: Sun, 2 Sep 2012 15:36:18 +0200 Subject: [PATCH 5/5] mp3dec: read Xing frame TOC index --- libavformat/mp3dec.c | 58 +++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 57 insertions(+), 1 deletion(-) diff --git a/libavformat/mp3dec.c b/libavformat/mp3dec.c index ffc1c35a60..7d0c2fb6ed 100644 --- a/libavformat/mp3dec.c +++ b/libavformat/mp3dec.c @@ -31,6 +31,9 @@ #define XING_FLAG_FRAMES 0x01 #define XING_FLAG_SIZE 0x02 +#define XING_FLAG_TOC 0x04 + +#define XING_TOC_COUNT 100 /* mp3 read */ @@ -94,6 +97,26 @@ static int mp3_read_probe(AVProbeData *p) //mpegps_mp3_unrecognized_format.mpg has max_frames=3 } +static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration) +{ + int i; + + if (!filesize && + !(filesize = avio_size(s->pb))) { + av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n"); + return; + } + + for (i = 0; i < XING_TOC_COUNT; i++) { + uint8_t b = avio_r8(s->pb); + + av_add_index_entry(s->streams[0], + av_rescale(b, filesize, 256), + av_rescale(i, duration, XING_TOC_COUNT), + 0, 0, AVINDEX_KEYFRAME); + } +} + /** * Try to find Xing/Info/VBRI tags and compute duration from info therein */ @@ -115,6 +138,8 @@ static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base) if(c.layer != 3) return -1; + spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */ + /* Check for Xing / Info tag */ avio_skip(s->pb, xing_offtbl[c.lsf == 1][c.nb_channels == 1]); v = avio_rb32(s->pb); @@ -124,6 +149,9 @@ static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base) frames = avio_rb32(s->pb); if(v & XING_FLAG_SIZE) size = avio_rb32(s->pb); + if (v & XING_FLAG_TOC && frames) + read_xing_toc(s, size, av_rescale_q(frames, (AVRational){spf, c.sample_rate}, + st->time_base)); } /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */ @@ -145,7 +173,6 @@ static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base) /* Skip the vbr tag frame */ avio_seek(s->pb, base + vbrtag_size, SEEK_SET); - spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */ if(frames) st->duration = av_rescale_q(frames, (AVRational){spf, c.sample_rate}, st->time_base); @@ -206,12 +233,41 @@ static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt) return ret; } +static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp, + int flags) +{ + AVIndexEntry *ie; + AVStream *st = s->streams[0]; + int64_t ret = av_index_search_timestamp(st, timestamp, flags); + uint32_t header = 0; + + if (ret < 0) + return ret; + + ie = &st->index_entries[ret]; + ret = avio_seek(s->pb, ie->pos, SEEK_SET); + if (ret < 0) + return ret; + + while (!s->pb->eof_reached) { + header = (header << 8) + avio_r8(s->pb); + if (ff_mpa_check_header(header) >= 0) { + ff_update_cur_dts(s, st, ie->timestamp); + ret = avio_seek(s->pb, -4, SEEK_CUR); + return (ret >= 0) ? 0 : ret; + } + } + + return AVERROR_EOF; +} + AVInputFormat ff_mp3_demuxer = { .name = "mp3", .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"), .read_probe = mp3_read_probe, .read_header = mp3_read_header, .read_packet = mp3_read_packet, + .read_seek = mp3_seek, .flags = AVFMT_GENERIC_INDEX, .extensions = "mp2,mp3,m2a", /* XXX: use probe */ };