[gst-devel] Gnonlin question: extract audio+video clip

[Olivier Aubert]

Hello

I am looking for a way to extract a video+audio clip from a video file.
The natural approach is to use gnonlin. I can extract the video with a
simple gst-launch command line:
gst-launch gnlfilesource name=video location=source.avi start=0 \
duration=5000000000 media-start=22000000000 media-duration=5000000000 ! \
identity single-segment=true ! progressreport update-freq=1 ! \
 ffmpegcolorspace ! theoraenc ! oggmux name=mux ! filesink location=/tmp/ex.ogv

However, combining video+audio is another matter. According to the
http://stackoverflow.com/questions/2481314/how-to-use-gnlcomposition-to-concatenate-video-files thread, I would need to use 2 gnlcompositions and  2 gnlfilesources to  achieve my goals. However, the following command line does not work:

gst-launch gnlfilesource name=video location=source.avi start=0 duration=5000000000 \
media-start=22000000000 media-duration=5000000000 ! identity single-segment=true \
! progressreport ! ffmpegcolorspace ! theoraenc ! oggmux name=mux \
! filesink location=/tmp/ex.ogv \
gnlfilesource name=audio location=source.avi start=0 duration=5000000000 \
media-start=22000000000 media-duration=5000000000 ! identity single-segment=true \
! vorbisenc ! queue ! mux.

and I think anyway that I should put a gnlcomposition in here, to keep
both streams in sync. Anyway, I feel that this is not achievable through
a gst-launch command-line, but would like to know if this feeling is
justified? Does anyone use a gst-launch command line to do such things?

Supplementary question: the remuxer.py from gst-python examples should
do this, but with some trickery (and I did not manage to get it to
successfully extract anything). Given its last real update dates back
from 2006, is it still considered as reference code?

Regards,
Olivier