- Word count: 564
- Average reading time: 2 minutes and 49 seconds (based on 200 WPM)
Somewhere in the middle of 2016
A very brief extension of youtube-dl that neatly downloads every (available) youtube video of a channel, coded in approx. 1 hour.
Youtube removes, for obvious reasons, a lot of videos. When a video is removed there is no way of finding out what the tile of a video was. I use youtube for several years already and have build a significant library over time. However when scrolling back through, for example, my favorites to figure out what I liked to listen to back then I noticed that so many titles were missing. I wanted to keep these titles, the audio I didn't care so much about but not knowing what the track's title was, was quite a let down.
To prevent this I looked a bit online for possible ways to download playlists of youtube and quickly stumbled upon youtube-dl. As the name, youtube-dl, suggests the purpose of this library was to rip youtube videos from a given URL, that url could be either a link to a video, playlist or a whole channel. However, when pointed to the direction of a channel it would download this into a single folder, bundling all the playlists together. I wanted to have each playlist in their respective folder.
Installing and setting youtube-dl up was a breeze, especially on Linux. However, YouTube uses multiple different audio formats:
.opus, .ogg, .webm and some others.
Initially only wanting a list of titles I figured: If I already have the audio, than why not?
For the pupose of conversion to a more widely used
.mp3 ffmpeg was suggested in the docs. Support for this is already integrated in youtube-dl so all I had to do was install ffmpeg and
hopefully it was able to find it automatically - no need for path editing - which was the case.
Having youtube-dl up and running I needed a command to download an entire channel. Such a command (for only audio) looks like this:
youtube-dl --extract-audio --ignore-errors -o "COMPLETE_FILE_PATH/%(title)s.%(ext)s" "COMPLETE_URL"
having the command ready, all I needed to do was recognize the channels I wanted to download, build a folder for it and generate the respective commands.
From the initial URL I requested the HTML, from this I extract the relevant playlist URLS and names, and with this list I let python generate 2 things:
- bash script with all the youtube-dl commands.
- folder structure in the main path.
Making this a lovely 2 command operation in which I first call the Python script and shortly after the freshly generated Bash script. I tried running the generated bash script from the python script to wrap it into a single command but the subprocess library threw some errors, which probably have something to do with a Python process calling a bash script which in its turn calls another Python process. I didn't look much into why this is, and at the moment of writing this I still don't exactly know why that happens but my suspicion goes to Python not allowing another Python interpreter to be ran inside a Python process, but then again, this is speculation and I'm not sure. It works as it is and I didn't want to spend too much time into it as my goal had already been achieved.