cur_dts in native timebase

Originally committed as revision 3153 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Michael Niedermayer 2004-05-23 21:36:23 +00:00
parent cdd5034f89
commit 77405fc866
3 changed files with 37 additions and 84 deletions

View File

@ -5,7 +5,7 @@
extern "C" {
#endif
#define LIBAVFORMAT_BUILD 4613
#define LIBAVFORMAT_BUILD 4614
#define LIBAVFORMAT_VERSION_INT FFMPEG_VERSION_INT
#define LIBAVFORMAT_VERSION FFMPEG_VERSION
@ -228,20 +228,13 @@ typedef struct AVStream {
seconds. */
int64_t duration;
/* the following are used for pts/dts unit conversion */
int64_t last_pkt_stream_pts;
int64_t last_pkt_stream_dts;
int64_t last_pkt_pts;
int64_t last_pkt_dts;
int last_pkt_pts_frac;
int last_pkt_dts_frac;
/* av_read_frame() support */
int need_parsing;
struct AVCodecParserContext *parser;
int64_t cur_dts;
int last_IP_duration;
int64_t last_IP_pts;
/* av_seek_frame() support */
AVIndexEntry *index_entries; /* only used if the format does not
support seeking natively */

View File

@ -163,7 +163,7 @@ int pcm_read_seek(AVFormatContext *s,
pos = (pos / block_align) * block_align;
/* recompute exact position */
st->cur_dts = (pos * AV_TIME_BASE) / byte_rate;
st->cur_dts = av_rescale(pos, st->time_base.den, byte_rate * (int64_t)st->time_base.num);
url_fseek(&s->pb, pos + s->data_offset, SEEK_SET);
return 0;
}

View File

@ -488,47 +488,6 @@ int av_read_packet(AVFormatContext *s, AVPacket *pkt)
/**********************************************************/
/* convert the packet time stamp units and handle wrapping. The
wrapping is handled by considering the next PTS/DTS as a delta to
the previous value. We handle the delta as a fraction to avoid any
rounding errors. */
static inline int64_t convert_timestamp_units(AVStream *s,
int64_t *plast_pkt_pts,
int *plast_pkt_pts_frac,
int64_t *plast_pkt_stream_pts,
int64_t pts)
{
int64_t stream_pts;
int64_t delta_pts;
int shift, pts_frac;
if (pts != AV_NOPTS_VALUE) {
stream_pts = pts;
if (*plast_pkt_stream_pts != AV_NOPTS_VALUE) {
shift = 64 - s->pts_wrap_bits;
delta_pts = ((stream_pts - *plast_pkt_stream_pts) << shift) >> shift;
/* XXX: overflow possible but very unlikely as it is a delta */
delta_pts = delta_pts * AV_TIME_BASE * s->time_base.num;
pts = *plast_pkt_pts + (delta_pts / s->time_base.den);
pts_frac = *plast_pkt_pts_frac + (delta_pts % s->time_base.den);
if (pts_frac >= s->time_base.den) {
pts_frac -= s->time_base.den;
pts++;
}
} else {
/* no previous pts, so no wrapping possible */
// pts = av_rescale(stream_pts, (int64_t)AV_TIME_BASE * s->time_base.num, s->time_base.den);
pts = (int64_t)(((double)stream_pts * AV_TIME_BASE * s->time_base.num) /
(double)s->time_base.den);
pts_frac = 0;
}
*plast_pkt_stream_pts = stream_pts;
*plast_pkt_pts = pts;
*plast_pkt_pts_frac = pts_frac;
}
return pts;
}
/* get the number of samples of an audio frame. Return (-1) if error */
static int get_audio_frame_size(AVCodecContext *enc, int size)
{
@ -598,15 +557,27 @@ static void compute_frame_duration(int *pnum, int *pden,
}
}
static int64_t lsb2full(int64_t lsb, int64_t last_ts, int lsb_bits){
int64_t mask = (1LL<<lsb_bits)-1;
int64_t delta= last_ts - mask/2;
return ((lsb - delta)&mask) + delta;
}
static void compute_pkt_fields(AVFormatContext *s, AVStream *st,
AVCodecParserContext *pc, AVPacket *pkt)
{
int num, den, presentation_delayed;
/* handle wrapping */
if(pkt->pts != AV_NOPTS_VALUE)
pkt->pts= lsb2full(pkt->pts, st->cur_dts, st->pts_wrap_bits);
if(pkt->dts != AV_NOPTS_VALUE)
pkt->dts= lsb2full(pkt->dts, st->cur_dts, st->pts_wrap_bits);
if (pkt->duration == 0) {
compute_frame_duration(&num, &den, s, st, pc, pkt);
if (den && num) {
pkt->duration = (num * (int64_t)AV_TIME_BASE) / den;
pkt->duration = av_rescale(1, num * (int64_t)st->time_base.den, den * (int64_t)st->time_base.num);
}
}
@ -621,6 +592,9 @@ static void compute_pkt_fields(AVFormatContext *s, AVStream *st,
st->codec.codec_id == CODEC_ID_H264) &&
pc && pc->pict_type != FF_B_TYPE)
presentation_delayed = 1;
/* this may be redundant, but it shouldnt hurt */
if(pkt->dts != AV_NOPTS_VALUE && pkt->pts != AV_NOPTS_VALUE && pkt->pts > pkt->dts)
presentation_delayed = 1;
}
/* interpolate PTS and DTS if they are not present */
@ -628,18 +602,22 @@ static void compute_pkt_fields(AVFormatContext *s, AVStream *st,
/* DTS = decompression time stamp */
/* PTS = presentation time stamp */
if (pkt->dts == AV_NOPTS_VALUE) {
pkt->dts = st->cur_dts;
/* if we know the last pts, use it */
if(st->last_IP_pts != AV_NOPTS_VALUE)
st->cur_dts = pkt->dts = st->last_IP_pts;
else
pkt->dts = st->cur_dts;
} else {
st->cur_dts = pkt->dts;
}
/* this is tricky: the dts must be incremented by the duration
of the frame we are displaying, i.e. the last I or P frame */
//FIXME / XXX this is wrong if duration is wrong
if (st->last_IP_duration == 0)
st->cur_dts += pkt->duration;
else
st->cur_dts += st->last_IP_duration;
st->last_IP_duration = pkt->duration;
st->last_IP_pts= pkt->pts;
/* cannot compute PTS if not present (we can compute it only
by knowing the futur */
} else {
@ -657,7 +635,6 @@ static void compute_pkt_fields(AVFormatContext *s, AVStream *st,
st->cur_dts = pkt->pts;
pkt->dts = pkt->pts;
}
//FIXME / XXX this will drift away from the exact solution
st->cur_dts += pkt->duration;
}
@ -678,6 +655,14 @@ static void compute_pkt_fields(AVFormatContext *s, AVStream *st,
}
}
/* convert the packet time stamp units */
if(pkt->pts != AV_NOPTS_VALUE)
pkt->pts = av_rescale(pkt->pts, AV_TIME_BASE * (int64_t)st->time_base.num, st->time_base.den);
if(pkt->dts != AV_NOPTS_VALUE)
pkt->dts = av_rescale(pkt->dts, AV_TIME_BASE * (int64_t)st->time_base.num, st->time_base.den);
/* duration field */
pkt->duration = av_rescale(pkt->duration, AV_TIME_BASE * (int64_t)st->time_base.num, st->time_base.den);
}
static void av_destruct_packet_nofree(AVPacket *pkt)
@ -751,29 +736,6 @@ static int av_read_frame_internal(AVFormatContext *s, AVPacket *pkt)
st = s->streams[s->cur_pkt.stream_index];
/* convert the packet time stamp units and handle wrapping */
s->cur_pkt.pts = convert_timestamp_units(st,
&st->last_pkt_pts, &st->last_pkt_pts_frac,
&st->last_pkt_stream_pts,
s->cur_pkt.pts);
s->cur_pkt.dts = convert_timestamp_units(st,
&st->last_pkt_dts, &st->last_pkt_dts_frac,
&st->last_pkt_stream_dts,
s->cur_pkt.dts);
#if 0
if (s->cur_pkt.stream_index == 0) {
if (s->cur_pkt.pts != AV_NOPTS_VALUE)
printf("PACKET pts=%0.3f\n",
(double)s->cur_pkt.pts / AV_TIME_BASE);
if (s->cur_pkt.dts != AV_NOPTS_VALUE)
printf("PACKET dts=%0.3f\n",
(double)s->cur_pkt.dts / AV_TIME_BASE);
}
#endif
/* duration field */
s->cur_pkt.duration = av_rescale(s->cur_pkt.duration, AV_TIME_BASE * (int64_t)st->time_base.num, st->time_base.den);
s->cur_st = st;
s->cur_ptr = s->cur_pkt.data;
s->cur_len = s->cur_pkt.size;
@ -881,6 +843,7 @@ static void av_read_frame_flush(AVFormatContext *s)
av_parser_close(st->parser);
st->parser = NULL;
}
st->last_IP_pts = AV_NOPTS_VALUE;
st->cur_dts = 0; /* we set the current DTS to an unspecified origin */
}
}
@ -1153,7 +1116,7 @@ av_log(s, AV_LOG_DEBUG, "%Ld %Ld %Ld / %Ld %Ld %Ld target:%Ld limit:%Ld start:%L
for(i = 0; i < s->nb_streams; i++) {
st = s->streams[i];
st->cur_dts = ts;
st->cur_dts = av_rescale(ts, st->time_base.den, AV_TIME_BASE*(int64_t)st->time_base.num);
}
return 0;
@ -1189,7 +1152,7 @@ static int av_seek_frame_generic(AVFormatContext *s,
for(i = 0; i < s->nb_streams; i++) {
st = s->streams[i];
st->cur_dts = timestamp;
st->cur_dts = av_rescale(timestamp, st->time_base.den, AV_TIME_BASE*(int64_t)st->time_base.num);
}
return 0;
@ -1804,10 +1767,7 @@ AVStream *av_new_stream(AVFormatContext *s, int id)
/* default pts settings is MPEG like */
av_set_pts_info(st, 33, 1, 90000);
st->last_pkt_pts = AV_NOPTS_VALUE;
st->last_pkt_dts = AV_NOPTS_VALUE;
st->last_pkt_stream_pts = AV_NOPTS_VALUE;
st->last_pkt_stream_dts = AV_NOPTS_VALUE;
st->last_IP_pts = AV_NOPTS_VALUE;
s->streams[s->nb_streams++] = st;
return st;