DSATube

A YouTube-style site made only of DSA videos: thousands of tutorials, lectures and interview walkthroughs, 1000+ playlists, related-video suggestions, and sharp thumbnails. React + Tailwind + shadcn/ui components. No API keys needed.

---

1. Install these first

Tool	Version	Check with	Download
Python	3.9+	python --version	python.org
Node.js	18+	node --version	nodejs.org
Git (only to deploy)	any	git --version	git-scm.com

If python isn't found, use python3 (and pip3) everywhere below.

2. Collect the data (the big step)

Open a terminal inside the folder, then:

pip install -r requirements.txt
python collect.py

It does three things:

1. Videos: about 1,800 YouTube searches (14 topics x many terms x 12 styles such as "tutorial", "in python", "in hindi"). Keeps DSA videos of 90 seconds to 12 hours, removes duplicates and live streams. Stops at 12,000 videos (--target).
2. Real YouTube playlists: finds up to 500 DSA playlists (--playlists) and saves the videos inside them.
3. Collections: builds 1000+ more playlists from your video list (by topic, language, channel, "for beginners", "interview questions", ...). These show as "DSATube collection" so you can tell them apart from real YouTube playlists.

It takes a long time (roughly 1-2 hours for everything). That's normal; you'll see progress lines. It is safe to stop: press Ctrl+C any time, progress is saved, and running python collect.py again continues where it stopped. For short sessions: python collect.py --max-minutes 20

Result: public/videos.json and public/playlists.json. The exact numbers depend on what YouTube returns. If you end with fewer than 1000 playlists, run it again (more videos means more collections).

3. Run the website

npm install
npm run dev

Open http://localhost:5173

4. Build the final site

npm run build

The finished website is the dist/ folder. Upload it to any static host (Netlify, Vercel, Cloudflare Pages, GitHub Pages).

5. Deploy free on GitHub Pages

1. Do step 2 on your computer first, so the data files are full.

2. Create an empty repository on github.com named dsatube.

3. In this folder run:

   git init
   git add .
   git commit -m "first commit"
   git branch -M main
   git remote add origin https://github.com/YOUR-USERNAME/dsatube.git
   git push -u origin main
   

4. GitHub: Settings > Pages > Build and deployment > Source: GitHub Actions.

5. Actions tab > "Update videos and deploy" > Run workflow.

6. In about 2 minutes: https://YOUR-USERNAME.github.io/dsatube/

Every day GitHub tries to add more data and redeploys. YouTube sometimes blocks GitHub's servers; then the site just keeps the data you uploaded. Re-run python collect.py on your own computer now and then and push the result.

6. Using the site

• Tabs: Home, Playlists, Watch later, Liked videos, Watch history. Liked, saved and history are stored in your browser.
• Home: topic buttons, length filter, search (press /), more videos load as you scroll.
• Watching: the player is on the left and 40 related videos are on the right (below the player on phones). Press the player's fullscreen button and it fills the screen on its own, with no suggestions.
• Playlists: each card opens a page listing every video inside, with Play all and Shuffle. While a playlist plays, the queue shows above the related videos with previous / next buttons.
• The browser Back button works everywhere and returns you to where you were.

7. About thumbnails

YouTube does not provide 4K thumbnails. The largest it offers is "maxres" (up to 1280x720), and DSATube always loads that one. If a video has no maxres thumbnail it automatically falls back to the next best size.

8. Options for collect.py

Option	Meaning
--target 20000	stop video search at this many videos (default 12000)
--playlists 800	real YouTube playlists to fetch (default 500, --skip-playlists to skip)
--pl-videos 100	max videos taken from one playlist (default 60)
--max-collections 2000	cap on generated collections (default 1500)
--per-query 100	results requested per search (default 60)
--max-minutes 30	stop after 30 minutes
--reset	forget finished searches and search everything again

Add favourite channels to CHANNELS and new search terms to TOPICS at the top of collect.py.

9. Troubleshooting

Problem	Fix
"No videos yet"	Run step 2, then refresh
Playlists tab empty	Run python collect.py (it builds playlists.json)
yt-dlp is missing	pip install -r requirements.txt
"YouTube returned nothing 6 times"	YouTube is rate-limiting you. Wait 30-60 min and run again (it resumes). Avoid VPNs
Few real playlists	Normal if YouTube changes its search page. pip install -U yt-dlp, run again. Collections still fill the tab
Collector seems stuck	Each search takes a few seconds. Ctrl+C is safe
npm install fails	Update Node.js to 18 or newer
Blank page after GitHub deploy	Wait 2 min, then hard refresh (Ctrl+Shift+R)
A video won't play	Its owner disabled embedding. Use the "YouTube" button on the watch page

10. Project layout

collect.py                 collector: videos, playlists, collections
requirements.txt           Python dependency (yt-dlp)
data/state.json            remembers finished searches so runs can resume
public/videos.json         video list (generated)
public/playlists.json      playlist list (generated)
src/App.jsx                routing, tabs, filters, infinite scroll
src/components/            WatchPage, PlaylistPage, cards, Thumb, ui/ (shadcn-style)
src/lib/related.js         "related videos" ranking
src/lib/library.js         Liked / Watch later / History (browser storage)
.github/workflows/         daily update + deploy