Short script using pytube and moviepy to automate the clipping of YouTube videos based on spreadsheets' timestamps (i.e. skip the first 6 hours of work in making highlights sourced from youtube!)

Input Structure

With csv's in the current directory containing the following columns:

• link (required)
  • YouTube link
• timestamp (required)
  • in the format MM:SS
• description
  • helpful blurb, part of which can be placed in output clip filename
• replay
  • whether or not to clip a larger length, in order to capture replay footage (leave blank for no)
• judgement (required)
  • "rating" of the clip, determining whether or not it is worth clipping
• comment
  • additional blurb (I often use to start pre-planning video structure)

Before running, at the top of clipper.py, adjust config constants to your needs.

For an example of input, the following sheet is an example that would work well with the script defaults.

hoyle.csv

link	timestamp	description	replay	judgement	comment
https://youtu.be/0T3LtBvANbg	3:50	6 pickup arm hit		clean
https://youtu.be/0T3LtBvANbg	4:10	straight fleche		nice
https://youtu.be/0T3LtBvANbg	18:25	the Novosjolov		perf
https://youtu.be/0T3LtBvANbg	21:25	hand pick to won		clean	sadly bad angle

Running

Using pipenv to run:

$ pipenv install
$ pipenv run python clipper.py

Debugging

Frequently, YouTube HTML page structure changes require a new version of pytube in order to scrape videos. If there seem to be errors from within the pytube library, try updating to the latest pytube release on pypi.

If there are still errors downloading videos and you have access to locally stored footage, you can also make changes to the how the queue of videos is constructed, and place the videos into individual folders (each within the video_output/<CSV_SHEET_NAME>/ folder by default).