We may use a thread that reads stderr in the background.Ī simpler solution is removing stderr=subprocess.PIPE, and adding '-hide_banner', '-loglevel', 'error' flags to FFmpeg (stderr is going to be written only in case of an error).įor debugging we may add -report flag (we may need it since we don't have the content that is written to stderr).įor testing we may build a synthetic pattern using FFmpeg CLI:įfmpeg -y -f lavfi -i testsrc=size=192x108:rate=10:duration=300 test.mp4Ĭode sample (playing from second 100 to 200): import subprocess When stderr buffer is full, FFmpeg sub-process halts and wait for stderr data to be read. The main issue is setting stderr=subprocess.PIPE without 'draining' the stderr pipe.

I used cv2.set but it took long time to play from start_second. I expect to play the video start from start_second and end at end_second. # Convert the frame buffer to a numpy arrayįrame = np.frombuffer(frame_buffer, dtype='uint8').reshape((height, width, 3))įrame_resized = cv2.resize(frame, (width, height), interpolation=cv2.INTER_AREA)Ĭv2.imshow('Video Player', frame_resized) # Read the frame from the subprocess stdout into the bufferīytes_read = (frame_buffer) Process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE) # Open a subprocess to execute the ffmpeg command '-t', str((end_seconds - start_seconds)), # Build the ffmpeg command to skip to the desired start time and end time import subprocessĭef play_video_with_ffmpeg(video_path, start_seconds, end_seconds):Ĭv2.namedWindow('Video Player', cv2.WINDOW_NORMAL)Ĭv2.moveWindow('Video Player', window_x, window_y) I want to play the video start at start_seconds and end at end_seconds. You can even use tmux itself for signalling:Īfter running this command, run and wait for tmux wait -L myscript Įdit your script or the 'source' command to run tmux wait -U myscript when finished.The following code stuck in the middle of the video. You can do the above using some other form of IPC (e.g. (This of course requires adding an 'echo' command at the end of the script, or after the 'source' command.) your script) actually writes something there. This will wait until another program (e.g.

You can create a named pipe and try to read from it. You can use inotify (Linux) or kqueue (FreeBSD) to watch the directory the kernel will tell you when the file shows up. You can poll (periodically check) whether the file has showed up. This is a poor method of running a script, but if you must use it, then your options are limited: It even doesn't know that those keypresses actually correspond to a script, much less know when that script is going to finish. You aren't telling tmux to run a command and wait you're telling it to simulate some keypresses. In fact, even tmux doesn't have this information.

Python subprocess background