29 lines
2.1 KiB
Markdown
29 lines
2.1 KiB
Markdown
# Step 2 - YouTube Music Authentication
|
|
|
|
🌐 **English** · [Dansk](Step-2-YouTube-Music-Authentication-da) · [Bahasa Melayu](Step-2-YouTube-Music-Authentication-ms)
|
|
|
|
## What the step does
|
|
|
|
YouTube Music has no simple API key. The `ytmusicapi` library works by borrowing the login of your browser: you copy the request headers of a logged-in Firefox session once (see [Setup and Run](Setup-and-Run)), and `ytmusicapi browser` turns them into `browser.json`. The program then opens `YTMusic("browser.json")` and counts your library playlists.
|
|
|
|
```bash
|
|
python -m musical_time_machine --check-login
|
|
```
|
|
|
|
On success it prints `Found N playlists in your library.` Without a valid `browser.json` it prints the renewal steps and exits with code 1, instead of a stack trace.
|
|
|
|
A live run with the author's `browser.json` (the output shows only the count, never the names of the playlists):
|
|
|
|
|
|
|
|
The screenshot shows the run **without** `browser.json` and the tests that use a fake YouTube Music.
|
|
|
|
|
|
|
|
## What we gained
|
|
|
|
- **How cookie-based authentication works.** Your browser proves who you are with cookies and headers on every request; copying them lets a script act as that session. That is also why `browser.json` is a secret, is ignored by git and expires after a while.
|
|
- **Hiding a library behind a port.** The program talks to a `MusicService` interface; only `infrastructure/ytmusic.py` imports `ytmusicapi`. The tests use a fake, and a different music service could replace it.
|
|
- **Failing helpfully.** A missing or expired file becomes a clear message that says how to renew it, because the error a user sees is part of the product.
|
|
- **A lazy connection.** The client is created on first use, so building the service never touches the network.
|