Re: [FFmpeg-devel] [PATCH 2/3] avformat/mxfdec: support subsecond precision of decoded timestamps

2018-12-25 Thread Marton Balint 



On Tue, 25 Dec 2018, Tomas Härdin wrote:


sön 2018-12-23 klockan 01:12 +0100 skrev Marton Balint:

> Signed-off-by: Marton Balint 
---
 libavformat/mxfdec.c | 5 -
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/libavformat/mxfdec.c b/libavformat/mxfdec.c
index d78f8ad2e4..0553adcb06 100644
--- a/libavformat/mxfdec.c
+++ b/libavformat/mxfdec.c
@@ -2556,12 +2556,14 @@ fail_and_free:
 static int64_t mxf_timestamp_to_int64(uint64_t timestamp)
 {
 struct tm time = { 0 };
+int msecs;
 time.tm_year = (timestamp >> 48) - 1900;
 time.tm_mon  = (timestamp >> 40 & 0xFF) - 1;
 time.tm_mday = (timestamp >> 32 & 0xFF);
 time.tm_hour = (timestamp >> 24 & 0xFF);
 time.tm_min  = (timestamp >> 16 & 0xFF);
 time.tm_sec  = (timestamp >> 8  & 0xFF);
+msecs= (timestamp & 0xFF) * 4;
 
 /* Clip values for legacy reasons. Maybe we should return error instead? */
 time.tm_mon  = av_clip(time.tm_mon,  0, 11);
@@ -2569,8 +2571,9 @@ static int64_t mxf_timestamp_to_int64(uint64_t timestamp)
 time.tm_hour = av_clip(time.tm_hour, 0, 23);
 time.tm_min  = av_clip(time.tm_min,  0, 59);
 time.tm_sec  = av_clip(time.tm_sec,  0, 59);
+msecs= av_clip(msecs, 0, 999);
 
-return (int64_t)av_timegm() * 100;
+return (int64_t)av_timegm() * 100 + msecs * 1000;


Looks OK


Thanks, pushed this and the other patch.



I kinda wonder how this and the muxer code could have ignored the
milliseconds. Did the old creation time metadata use seconds only?


Yes, it used to be in this format: "%Y-%m-%d %H:%M:%S".

Regards,
Marton
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH 2/3] avformat/mxfdec: support subsecond precision of decoded timestamps

2018-12-25 Thread Tomas Härdin 
sön 2018-12-23 klockan 01:12 +0100 skrev Marton Balint:
> > Signed-off-by: Marton Balint 
> ---
>  libavformat/mxfdec.c | 5 -
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/libavformat/mxfdec.c b/libavformat/mxfdec.c
> index d78f8ad2e4..0553adcb06 100644
> --- a/libavformat/mxfdec.c
> +++ b/libavformat/mxfdec.c
> @@ -2556,12 +2556,14 @@ fail_and_free:
>  static int64_t mxf_timestamp_to_int64(uint64_t timestamp)
>  {
>  struct tm time = { 0 };
> +int msecs;
>  time.tm_year = (timestamp >> 48) - 1900;
>  time.tm_mon  = (timestamp >> 40 & 0xFF) - 1;
>  time.tm_mday = (timestamp >> 32 & 0xFF);
>  time.tm_hour = (timestamp >> 24 & 0xFF);
>  time.tm_min  = (timestamp >> 16 & 0xFF);
>  time.tm_sec  = (timestamp >> 8  & 0xFF);
> +msecs= (timestamp & 0xFF) * 4;
>  
>  /* Clip values for legacy reasons. Maybe we should return error instead? 
> */
>  time.tm_mon  = av_clip(time.tm_mon,  0, 11);
> @@ -2569,8 +2571,9 @@ static int64_t mxf_timestamp_to_int64(uint64_t 
> timestamp)
>  time.tm_hour = av_clip(time.tm_hour, 0, 23);
>  time.tm_min  = av_clip(time.tm_min,  0, 59);
>  time.tm_sec  = av_clip(time.tm_sec,  0, 59);
> +msecs= av_clip(msecs, 0, 999);
>  
> -return (int64_t)av_timegm() * 100;
> +return (int64_t)av_timegm() * 100 + msecs * 1000;

Looks OK

I kinda wonder how this and the muxer code could have ignored the
milliseconds. Did the old creation time metadata use seconds only?

/Tomas
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH 2/3] avformat/mxfdec: support subsecond precision of decoded timestamps

2018-12-22 Thread Marton Balint 
Signed-off-by: Marton Balint 
---
 libavformat/mxfdec.c | 5 -
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/libavformat/mxfdec.c b/libavformat/mxfdec.c
index d78f8ad2e4..0553adcb06 100644
--- a/libavformat/mxfdec.c
+++ b/libavformat/mxfdec.c
@@ -2556,12 +2556,14 @@ fail_and_free:
 static int64_t mxf_timestamp_to_int64(uint64_t timestamp)
 {
 struct tm time = { 0 };
+int msecs;
 time.tm_year = (timestamp >> 48) - 1900;
 time.tm_mon  = (timestamp >> 40 & 0xFF) - 1;
 time.tm_mday = (timestamp >> 32 & 0xFF);
 time.tm_hour = (timestamp >> 24 & 0xFF);
 time.tm_min  = (timestamp >> 16 & 0xFF);
 time.tm_sec  = (timestamp >> 8  & 0xFF);
+msecs= (timestamp & 0xFF) * 4;
 
 /* Clip values for legacy reasons. Maybe we should return error instead? */
 time.tm_mon  = av_clip(time.tm_mon,  0, 11);
@@ -2569,8 +2571,9 @@ static int64_t mxf_timestamp_to_int64(uint64_t timestamp)
 time.tm_hour = av_clip(time.tm_hour, 0, 23);
 time.tm_min  = av_clip(time.tm_min,  0, 59);
 time.tm_sec  = av_clip(time.tm_sec,  0, 59);
+msecs= av_clip(msecs, 0, 999);
 
-return (int64_t)av_timegm() * 100;
+return (int64_t)av_timegm() * 100 + msecs * 1000;
 }
 
 #define SET_STR_METADATA(pb, name, str) do { \
-- 
2.16.4

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel