Some YouTube content creators split their videos into chapters. A subset of them kindly offer the list of chapters titles on description video, with their starting timestamp. Sometimes they decide not to show them.
How can I get the list of chapters titles?
One-liner to do this using yt-dlp (better alternative to youtube-dl) and jq JSON processor:
yt-dlp --dump-json videoIDorURL | jq --raw-output ".chapters[].title" To get starting timestamps:
| jq --raw-output ".chapters[].start_time" | awk '{printf("%d:%02d:%02d\n",($1/60/60%24),($1/60%60),($1%60))}' With paste and process substitution you can combine both. This is a function you could add to your .bashrc/.zshrc:
function get_chapters_times() { paste <(yt-dlp --dump-json $1 | jq --raw-output ".chapters[].start_time" | awk '{printf("%d:%02d:%02d\n",($1/60/60%24),($1/60%60),($1%60))}') <(yt-dlp --dump-json $1 | jq --raw-output ".chapters[].title") } Will return:
$ get_chapters_times https://youtu.be/DxL2HoqLbyA 0:00:00 Intro 0:02:15 History 0:04:16 Ideal Engine 0:09:48 Entropy 0:11:03 Energy Spread 0:14:49 Air Conditioning 0:17:26 Life on Earth 0:19:35 The Past Hypothesis 0:21:43 Hawking Radiation 0:23:31 Heat Death of the Universe 0:24:52 Conclusion Use yt-dlp --split-chapters to download each chapter.
In python, you can do this:
import datetime import yt_dlp ydl = yt_dlp.YoutubeDL({'dumpjson':True } ) dnwl_setting = False with ydl: result = ydl.extract_info( ytdl_variables.video_url, download=dnwl_setting ) for chpt in result['chapters'] : str_start_time = chpt['start_time'] str_start_time_cnv = str(datetime.timedelta(seconds=str_start_time) ) str_title = chpt['title'] # for youtube timecodes print ( f"time = {str_start_time_cnv} | title = {str_title } ") Here's a fancier one-liner inspired by Pablo's answer that does away with process substitution and having to define functions:
yt-dlp videoURL --dump-json | jq -r ".chapters[] | {start_time, title}" | awk -v count="2" '++count==4{$0=sprintf(" \"start\":%d:%02d:%02d",($2/60/60%24),($2/60%60),($2%60));count=0} 1' | awk -F '":' '{print $2}' | gsed -r -e '/^\s*$/d' -e 's/^ "|"$//g' | paste -d " " - -
The awks and gsed could probably be merged into each other at the cost of more complexity.
If using yt-dlp, you may also benefit from --embed-chapters (alias: --add-chapters) argument.
